Linking Smallholder Farmers to the Heilongjiang Province Crop Rotation Project: Assessing the Impact on Production and Well-Being

Abstract

Food security and environmental protection—led by sustainable agricultural development—are key development goals of Heilongjiang Province. One of the main challenges facing the Heilongjiang Province is improving soil and livelihood by integrating smallholder farmers into the pilot crop rotation project. This paper investigates a comprehensive project—the new crop rotation pilot project in Heilongjiang (NCRPPH)—which aims to improve the livelihood of the pilot participants by involving smallholder farmers in the pilot crop rotation program and connecting them with the food industry through farmers’ cooperatives. This paper analyzes the impact of the NCRPPH on farmers’ crop rotation, grain yield, multi-agent cooperation, food security, and education based on data collected in 2019 and other retrospective information. The instrumental variable method and three different estimation strategies are used to solve the endogenous problem. The results show that the project has a vast and positive impact on the gross and net value of grain production per hectare and the share of products sold to food manufacturers through cooperatives. Regardless of the farm size, farmers have equal opportunities from which they can benefit. In addition, our analysis shows that the NCRPPH improves the educational situation and food security without affecting crop rotation practices. Finally, our research proves the effectiveness of this project.

1. Introduction

Heilongjiang Province, with its vast expanse of black land, produces 10% of China’s grain. However, as a result of international grain trade and domestic collection and storage policies, corn planting area has increased sharply, soybean planting area has decreased sharply, wheat has almost vanished, and the original rotation mode has been destroyed, with black soil fertility overdraft and cultivated land showing degradation [1]. In May 2016, the central government deliberated and approved a pilot scheme for exploring the implementation of a crop rotation and fallow system, focusing on pilot rotation in the cold and cool areas of northeast China. The key to the pilot scheme is to develop a scientific and acceptable rotation system [2]. Furthermore, agriculture in the black soil region of northeast China is undergoing substantial transition and development. Traditional farmers are increasingly becoming more differentiated by the new crop rotation method, and a new agricultural production and management system is forming [3]. Many agricultural production management subjects must cooperate in the new crop rotation system since they have diverse interests and decision-making systems [4]. Crop rotation may also have a significant impact on the livelihoods of small farmers, raising concerns regarding national food security [5].

In a relatively new collaborative chain (CC), creating systematic links between and among small-scale farmers, cooperatives, food processors, and consumers is not straightforward. To address this predicament, small-scale farmers and other participants in the collaborative chain must pay attention to cereal rotation mode—coordination of the agricultural products market. This includes creating institutional arrangements for connecting farmers to one another and to marketing outlets and the introduction of research, extension, and financial institutions [6,7,8]. Studies have shown that contract farming and farmer cooperatives are two common institutional arrangements in Heilongjiang [9,10,11,12,13].

Contract farming is a mode in which farmers produce according to the quality, price and quantity agreed with processors or intermediaries. It is a major component of Heilongjiang’s Growth and Reform Strategy II to bring small farmers and sustainable markets together in order to promote agricultural development. In that, small farmers tend to enjoy improved market prices and easier access to credit and technical assistance [14,15,16]. It may also assist in reducing marketing risk by assuring more consistent pricing than what is available on the open market [17]. The final effect might be an increase in pricing for producers and/or an increase in the amount of product sold, resulting in increased earnings for the producers.

However, there are also some problems in which local inequality will grow as contract agriculture becomes more accessible to affluent farmers with the required resources and skills [15]. Increased manufacturing and marketing risks, as well as potential power imbalances, may have negative consequences [18]. Recent research has investigated the possible consequences of contract agriculture [19,20,21,22]. In a previous study, Rao and Qaim, using a large data set including different crops, companies, and areas in Madagascar, discovered that contract sales to supermarkets in Kenya increased vegetable growers’ income [22]. Bellemare determined that contract agriculture resulted in a significant increase in income [23]. According to Warning and Key, Senegalese farmers’ revenue from contract peanut production increased dramatically. Two studies on contract agriculture have been published in China, one on watermelon by Miyata and the other on apples and scallions by Ito (2012). It was found that contract agriculture has a positive effect on both food security and revenue [24,25,26].

Ton et al. conducted a thorough evaluation of 26 contractual agricultural arrangements in 13 developing countries and found that they had positive income benefits, with a total income effect of 38% [27]. Even though just two of the evaluated research identified negative income effects in some contract agricultural scenarios [20], Ton et al. discovered significant variances, depending on contract type, crop, and institutional environment [27]. Ragasa et al. (2018) found that while contractual maize planting in Ghana resulted in increased technology adoption and yield, it did not boost farm income. Such negative or minor impacts [18], according to Ton et al. [27], may be routinely understated due to publishing and other biases and should be thoroughly reviewed further. They discovered, for example, that contract farmers were, on average, wealthier than conventional farmers in the region in terms of land and other assets [28,29].

Farmer cooperatives are part of the Heilongjiang Government’s strategy for increasing agricultural production. Farmers’ cooperatives, as the driving force behind the new type of management, can help connect farmers to markets and alleviate poverty [30,31,32,33]. Marketing cooperatives can help small-scale farmers overcome supply constraints, such as minimum acquisition volume, quality requirements, and frequency, allowing them to participate in high-value markets and contract agriculture programs [17]. In general, the combination of farmers’ production activities can effectively reduce the additional expenditure and diseconomies of scale in the production and trading process. It may also assist organizations to increase their negotiating power and obtain access to information, as well as sign contracts with buyers that require a high number of simultaneous transactions. On the other hand, cooperatives may be used to empower rural elites and create order by concentrating market forces [34].

While there are limited studies on the income benefits of cooperatives in Heilongjiang, grain marketing performance has been investigated in various studies with varied results. Bernard et al. found no average impact of food cooperatives on agricultural commercialization or disadvantaged farmers; however, they did discover an impact on prices, implying some beneficial bargaining power [35]. Bernard and Spielman discovered that the poorest farmers are rarely included in food marketing cooperatives, which prevents them from reaping the benefits of high prices [36]. Francesconi and Heerink discovered that cooperative members are highly marketed, but only in marketing cooperatives [34]. They discovered that as cooperatives diversify their activities beyond marketing, their ability to provide marketing services suffers [37]. However, in some regions, cooperatives do not assist small farmers. Moslem and Mohammad found that small farmers in Iran do not require agricultural cooperatives to provide cold storage services and sales channels in response to drought [38]. Therefore, we doubt the effectiveness of agricultural cooperatives in Heilongjiang Province.

In order to evaluate the effect of small farmers’ participation in CC, this paper focuses on the impact of the New Crop Rotation Pilot Project in Heilongjiang (NCRPPH) on output and other well-being factors. Moreover, in this paper, we propose three priorities for evaluation. The first focuses on the impact of the NCRPPH on grain production, the second measures the contribution of the NCRPPH in agricultural product marketing, and the third is used to evaluate the changes in the NCRPPH to the following two important well-being factors: education and nutrition.

Similarly to the evaluation of most large-scale agricultural cooperation chain projects, this paper utilizes large-scale cross-sectional data and preliminary information collected in 2019 to increase the credibility of the evaluation results. We selected instrumental variable (IV) to evaluate the effectiveness of the NCRPPH, and three widely used estimation strategies were implemented to ensure the credibility of the results.

The rest of the article is set out as follows: Section 2 briefly introduces the crop rotation policies concerning Heilongjiang. Section 3 explains the NCRPPH project and the theory of our evaluation frames. Section 4 discusses the data and methodology. Finally, Section 5 and 6 discuss the empirical results and policy implications, respectively.

2. Crop Rotation Policy in Heilongjiang

China’s Heilongjiang Province is a major grain-producing region. Over the past 12 years, the overall grain output has increased and is expected to exceed 75 billion kg in 2021, contributing significantly to national food security. However, it cannot be denied that every breakthrough in the process of raising grain output in Heilongjiang Province in recent years has been achieved by increasing the planting area and increasing the intake of chemical fertilizers and pesticides. In Heilongjiang Province, the proportion of crops that play a major role in the stability of the farming system is currently out of balance. The single cropping system and variety problem are becoming increasingly significant [39,40]. As a result, it is critical to change the current planting structure in Heilongjiang Province, implement substitution planting and crop rotation, and establish a new ecological and efficient farming system, either from the perspective of sustainable agricultural development or the protection of national food security [41,42,43,44].

Table 1. Grain sowing area and percentages of different regions of Heilongjiang Province (2014).

Region	Crop		Rice		Wheat		Corn		Soybean		Potato
	Area (ha)	Proportion	Area (ha)	Proportion	Area (ha)	Proportion	Area (ha)	Proportion	Area (ha)	Proportion	Area (ha)	Proportion
Whole	1477.53	100	399.67	27.05	12.27	0.83	664.20	44.95	314.60	21.29	16.53	1.12
Harbin	204.40	13.83	63.80	4.32	0.00	0.00	110.47	7.48	17.87	1.21	1.95	0.13
Qiqihar	229.40	15.53	32.07	2.17	0.02	0.00	129.40	8.76	52.60	3.56	8.27	0.56
Jixi	49.27	3.33	17.33	1.17	0.01	0.00	25.00	1.69	5.60	0.38	0.10	0.01
Hegang	20.40	1.38	11.73	0.79	0.00	0.00	4.60	0.31	3.73	0.25	0.03	0.00
Shuangyashan	42.07	2.85	8.53	0.58	0.00	0.00	24.60	1.66	6.93	0.47	0.07	0.00
Daqing	75.73	5.13	10.40	0.70	0.22	0.01	53.40	3.61	2.07	0.14	0.15	0.01
Yichun	23.93	1.62	3.87	0.26	0.00	0.00	4.53	0.31	14.27	0.97	0.07	0.00
Jiamusi	130.47	8.83	57.87	3.92	0.09	0.01	38.00	2.57	30.07	2.04	0.69	0.05
Qitaihe	17.80	1.20	1.87	0.13	0.00	0.00	12.27	0.83	2.00	0.14	0.15	0.01
Mudanjiang	65.07	4.40	4.67	0.32	0.11	0.01	30.13	2.04	19.27	1.30	0.55	0.04
Heihe	123.27	8.34	2.53	0.17	10.05	0.68	32.73	2.22	70.67	4.78	0.53	0.04
Suihua	190.87	12.92	35.07	2.37	0.00	0.00	115.73	7.83	34.33	2.32	1.61	0.11
Daxinganling	15.13	1.02	0.00	0.00	1.04	0.07	0.40	0.03	14.80	1.00	0.35	0.02
Nongken	286.27	19.37	150.07	10.16	0.76	0.05	82.93	5.61	43.60	2.95	2.00	0.14

Source: authors.

shows that the overall sown area of crops in Heilongjiang Province in 2014 was 0.148 billion ha, with the Songnen Plain and Sanjiang Plain accounting for the majority of this area. The Songnen Plain is Heilongjiang Province’s largest planting region. Crops sown in Harbin, Qiqihar, Suihua, Daqing, and other regions (excluding the General Administration of Land Reclamation) accounted for 47.40% of the province’s total crop sown area of 0.07 billion ha, with corn accounting for 58%, rice 20%, and soybeans 15%. In the Sanjiang Plain, crop sown area in Jiamusi, Shuangyashan, Jixi, Hegang, Qitaihe, and other regions accounted for 17.60% of the province’s total, or about 2.6 million ha (excluding the General Administration of Land Reclamation), with corn accounting for 40%, rice 37%, and soybeans 19%. The seeded area of agricultural crops in Daxinganling, Heihe, and Yichun amounts to 10.98% of the province’s total, or roughly 162,300 hectares (excluding the General Administration of Land Reclamation), with soybeans accounting for 61%, corn 23%, and wheat 3%—accounting for 7% of the total. In summary, rice, corn, and soybeans are currently the most important crops in Heilongjiang Province. According to data, Heilongjiang Province has more than 4 million ha of continuous corn cropping, and Heihe has more than 133,300 ha of continuous soybean farming. Corn–soybean rotation and corn–soybean–wheat rotation have both vanished. The current cropping system and single species problem in Heilongjiang Province has been extremely serious, and the farming system’s stability has deteriorated, resulting in a decrease in the farmland ecosystem’s resistance to natural disasters, as well as a significant increase in disaster risks and losses.

For many years, Heilongjiang Province’s continuous staple crop agricultural area has grown, resulting in decreased crop yields. Moreover, heavy crop yield loss—regarding staple crops—is becoming a serious problem as continuous monocropping is extended, resulting in a reduction in agricultural production. According to long-term positioning test observations at the Chinese Academy of Sciences Hailun Agroecological Experiment Station, Soybean yields dropped by 9% after two years of continuous planting and by up to 35% after five years of continuous planting. The longest continuous crop time in northeast China’s soybean production regions is up to 30 years.

Continuous crops from staple crops—not regarding the importance of fertilizer crop rotation, organic fertilizer and straw return to the field, and other fertilization—results in a soil nutrient imbalance, a decline in the quality of the cultivated layer of land, and a reduction in sustainable production capacity. According to survey data, the average depth of soil tillage in Heilongjiang Province is barely 15 cm, making it neither drought- nor flood-resistant, and the ability to hold water and fertilizer has significantly decreased. Farmers rely excessively on chemical fertilizers to sustain production, and large amounts of applied chemical fertilizers not only raise the expense of agriculture but also pollute farmland’s surface. The cultivated layer’s soil organic matter level is currently 1.5% to 2.5% in Heilongjiang Province, and the quality of black soil has deteriorated dramatically.

Continuous cropping of staple grain crops not only affects crop production, but also exacerbates the challenges associated with continuous croppings, such as diseases, pests, and weeds, lowering fertilizer and other material input transformation efficiency. According to the observations of the Chinese Academy of Sciences’ agroecological experimental station, nitrogen usage efficiency has declined from 31% to 25% in the last 20 years, and soil natural productivity has decreased from 89% to 53%. In the practice of agricultural production, farmers will raise their investment in pesticides, chemical fertilizers, and agronomic methods in order to overcome the barriers of continuous cropping and sustain crop output, hence increasing the cost of crop production.

A reasonable crop rotation, according to some researchers [45,46], is a combination of use land and use nutrients that can change the physical and chemical properties of the soil, improve the soil structure, regulate the supply of nitrogen, phosphorus, potassium, and other mineral nutrients required by crops, as well as facilitate the control of pests and diseases, particularly crop-associated weeds and parasitic weeds, and many other benefits. For example, in Heilongjiang Province’s Heihe area, maize production is 8000 kg/ha. When computed at 1.9 Renminbi (RMB)/kg, the output value is 12,800 RMB/hectare. After subtracting the 7400 RMB/ha production cost, the gain is 5400 RMB/ha. After subtracting the production cost of 4000 RMB/ha, the hay yield (after one year) is 13,500 kg/ha, the output value is 10,800/ha, and the benefit is 6800 RMB/ha when alfalfa is produced on land with the same fertility. In light of the low benefits of grain planting and the impact of the global grain market, and in light of the ecological, regional, and agricultural structural characteristics of the Songnen Plain and Sanjiang Plain, it is important to accelerate the adjustment of planting structure in Heilongjiang Province, establish distinct modern farming system models, and build an agricultural system technology system with comprehensive high and stable crop yields and sustainable agricultural development. In addition, rotation farming should cooperate with other conservation tillage measures (e.g., less-tillage, no-tillage, furrow and ridge farming, stubble mulching, straw mulching, reasonable dense planting, farmland shelter forest construction, etc.) to jointly repair the quality and ecological problems of black soil, improve the ability of black soil to resist soil erosion and natural disasters, and improve the comprehensive production capacity of black soil.

Historically, the crop rotation area and other cereal production have been an important policy goal of the Heilongjiang Government [47]. For example, the National Soybean Research Program (NSRP) was implemented to develop fundamental scientific knowledge and to apply rotation technology in order to boost and sustain soybean output in Heilongjiang [48]. In February 1993, restrictions on private transactions were lifted in the grain market, and the monopoly position of the official pricing and Marketing Committee became a feature of the past. However, according to Xiong [49], the market remained “efficient but poor” because of limitations.

Several development policies in Heilongjiang have made crop rotation important, including the Rural Revitalization and Agricultural Supply-Side Structural Reform, which both aim to boost industrial development and food security. In order to grow and transform agriculture, it is necessary to strengthen the commercialization of small-scale farmers and links with CC.

Furthermore, the Heilongjiang Government has set up the farmer cooperative economic guidance division to enhance CC as a strategic sub-sector that contributes to sustainable agriculture and food security. Moreover, the multi-agent agricultural growth program is designed to help farm cooperative unions and agribusinesses expand crop rotation areas, increase crop productivity, and increase local supply. Meanwhile, agricultural cooperatives play a significant role in increasing production and commercializing cereals [50,51,52,53,54,55,56].

3. New Crop Rotation Pilot Project in Heilongjiang (NCRPPH)

The NCRPPH was presided over by several Heilongjiang stakeholders as part of the “Exploring the Pilot Scheme of Implementing the Crop rotation and Fallow System” framework, between 2016 and 2020. It was based on the Northeast Revitalization Project widely implemented for many years in the region, specifically the rural vitalization project, which involved a variety of different local actors, such as cooperatives, agricultural universities, extension stations, and government.

Moreover, its goal was to boost smallholder cereal production and marketing by enhancing the quality of farmed land and crop varieties, developing cooperatives, and creating direct linkages between cooperatives and Heilongjiang food producers in Harbin via contract farming agreements. Later, the northeastern region of Heilongjiang was designated as a cultivated land rotation pilot area [57,58].

The project is located in the grain production area of Heilongjiang. Crop rotation pilot work will be carried out in the main traditional soybean producing areas of Heihe, Yichun, and in other counties (cities, districts) in northern Heilongjiang Province, as well as in the main traditional soybean producing areas of the Nongken Jiu San and Bei’an administrations. The area where the project is located is flat, with an average altitude of about 850 m, the latitude is between 47–51 degrees north latitude, and the socioeconomic conditions and natural conditions (soil, temperature, and precipitation) are roughly the same [57,59].

3.1. Project Objectives and Characteristics

The NCRPPH’s general aim is to reduce the use of chemical fertilizers and pesticides, efficiently use land resources, adjust planting structure, relieve pressure on land, promote the sustainable development of ecological resources and raise the quality of agricultural products, meet people’s demand for diversified agricultural products, and also increase farmers’ financial income. Its strategy revolves around making it easier for smallholders to join the agricultural co-op chain as part of a rotation pilot program. However, farmers have not achieved satisfactory results due to production and coordination issues (

Table 2. NCRPPH’s breakthrough in the CC of crop rotation.

Breaking Point	Actions	Participating Actors
Small-sized farms (and small per-farm production)	Creation of cooperative’s unions	Cooperatives
	Cooperatives’ strengthening	Associations
	New storage facilities at the cooperative/town level	Farmers
		Institutions at the regional and zonal levels
Farmers’ risk awareness and avoidance ability	Contract agriculture is the link between farmers and the market Peer learning Pricing power	Cooperatives and Associations
		Food producers
		Institutions Regional and Zonal
		Provincial Center for Agricultural Research (quality lab checks)
Inadequate coordination among participants in the CC	Exchange of expertise and education of farmers and cooperative managers	Unions & Cooperatives Food producers
	Increasing the strength and participation of Local management agencies in the CC	Institutions Regional and Zonal
High workload of production (i.e., change in production rhythm)	Enhance the technical and management capacity of cooperatives	International consultants
	Developing new seed varieties with adapted rotation plans (patent-pend by research institutions) Establish a cooperative chain of Agricultural technology institutions and crop rotation coordination cooperatives Training on appropriate crop rotation Give farmers technical guidance on fertilization and appropriate rotation practices	Bei’an Agricultural Technology Extension Center Farmers Select farmers who are committed to the seed rotation cycle The NCRPPH program directly via cooperatives Cold land Agricultural Advanced Technology Research Center (CLAATRC)

Source: authors.

).

In 2016, the NCRPPH launched an array of interconnected activities addressing the following two main fields: the first focused on the technical side of production, such as implementing appropriate agronomic practices (e.g., straw returning and soil testing formula fertilization technology), introducing crop rotation varieties, and providing key assets to cooperatives. The second was the overall institutional structure of the CC, which focused on capacity-building among cooperatives, tying cooperatives to publicly funded agricultural research centers, and establishing cooperative plans for contract farming, something that has not been carried out in the region previously. These arrangements involve signing contracts among farmers, cooperatives, and food processing enterprises to ensure that they comply with the quality standards stipulated in the contract.

The NCRPPH was expected to result in a shift in the CC’s power dynamics. Farmers in neighborhood markets are frequently harmed by traders, intermediaries, and lenders’ increased bargaining power [50,51]. It was estimated that as a link between farmers and food manufacturing enterprises, cooperatives are conducive to strengthening farmers’ negotiating power. According to data obtained by the Cold Land Agricultural Advanced Technology Research Center (CLAATRC) in the NCRPPH area—which annually pilots around 860,000 ha—the initiative collaborated with more than 100 cooperatives.

3.2. Rationale and Framework

The NCRPPH uses both improvements in the CC’s institutional architecture and technical improvements to upgrade small-scale crop rotation production. A key part of the NCRPPH’s effectiveness rests in the direction of how these approaches interact with each other, since impacts at the farm level are only possible with collective action. In a sound rotation production cycle, farmers, cooperatives, and public research centers can use improved seeds. Furthermore, cereal quality is measured at the cooperative/regional level; therefore, if other members of the cooperative do the same, the farmer will only be incentivized to invest in quality activities—such as improved training, highly efficient processing, and effective crop rotation.

Farmers’ bargaining strength is likely to increase as they acquire direct access to the national market, particularly through cooperatives. Additionally, alternative marketing channels provided by cooperatives may stimulate competition by compelling other middlemen to offer competitive rates. Moreover, the transformation brought about by improved access to national markets may enable smallholder farmers to produce higher-value crops, allowing farmers to retain a larger part of added value. It is projected that these changes to production and market systems will increase farmers’ grain revenues by increasing grain prices and/or output and sales quantities. As long as additional expenditures are less than the rise in household income, household income may increase. By raising family income, the project has the potential to improve other aspects of farmers’ welfare, including food safety, education, and health (Figure 1).

However, the NCRPPH may also bring unexpected negative effects. For example, higher yield crops will reduce crop rotation or reduce the planting area of traditional crops. In addition, natural disasters including droughts, floods, and typhoons will also affect the effective implementation of the NCRPPH. Policy continuity and population mobility in the area might also gradually decrease the NCRPPH’s efforts to promote a positive business environment and build relationships across various levels [52]. The NCRPPH’s efforts could also be hampered by price instability within the national and international grain markets since Heilongjiang is still heavily dependent on foreign fertilizer imports.

4. Methodology

In 2019, a large-scale household survey was conducted to assess the project’s impact. Following the survey, the data were analyzed using a variety of econometric techniques. In order to ensure the representativeness of the sample data, this study conducted in-depth interviews with stakeholders such as NCRPPH employees and researchers from the CLAATRC, developed a questionnaire, determined the participating cooperatives and farmers (treatment), and developed a sample design.

4.1. Sample Groping and Data Collection

All farmers were divided into seven groups according to their living areas, whether they participate in cooperatives, and whether they participate in the NCRPPH. Figure 2 clearly shows the relationship between the seven groups of farmers.

Group T-a, Group T-b, and Group C-a are from cooperative farmers participating in the NCRPPH. Group T-a are treated farmers who decided to rotate their produce and sell it directly to food manufacturers by providing it to cooperatives. Their relative incidence varies by cooperative and district (i.e., it approaches 100% in Bei’an, but is less than that in other districts). Farmers in Group T-b are part of the rotation experiment but sold their crops through other channels. Because the impact of other possible prices cannot be excluded, they are also part of the treated group. Group C-a consists of farmers who belong to one of the cooperatives involved in the NCRPPH but did not consent to participate in the project. Because there was no difference in natural conditions and grain cultivation between this group and the treatment group, it can be used as the control group.

Group C-b and the other three groups are from cooperatives that do not participate in NCRPPH and farmers that do not participate in cooperatives. Farmers in Group C-b are members of a non-treated cooperative in a non-treated town. If the town is similar to the treated town in terms of agricultural ecology, social situations, and demonstrative visuals, and is about to participate in the NCRPPH, it might be chosen as a control group. Farmers in Group N-a are members of a non-treated cooperative in a treated town. There are just a few communities with more than one cooperative. This group was omitted from the control group because of the exceedingly significant danger of treatment trickle-down and overflow. Farmers who are not formally members of cooperatives, but live in a treated town, comprise Group N-b. This group cannot be utilized as a control group because these farmers are difficult to identify. Finally, non-cooperative farmers living in a non-treated town make Group N-c an inadequate control group for the research.

Ten out of the twenty-five NCRPPH cooperatives (by the time the poll was conducted, fifteen people had already departed the program) and nine out of twenty-eight non-treated cooperatives were picked from the total. The non-treated cooperatives are chosen based on the following criteria: these cooperatives that were part of the previous rural vitalization project; the agroecological characteristics here are similar to those of the treatment group; they were located far away from the treated cooperatives to avoid spillover effects; they did not experience extreme climate shock during the reference cropping season.

An ex ante power analysis was calculated using the following two variables from the CLAATRC statistics: the number of small farmers farming soybeans, rice, and corn in the area (95.3%), and the proportion of small farmers not participating in cooperatives (96.2%). Based on these two data, the number of households that were required to be interviewed was 734 and 614, respectively, with a 1% margin of error (183 and 154 with 2% error). Finally, based on the power analysis, a total of 733 farmers from 19 cooperatives were selected from 53 cooperatives to complete the interview (35–45 farmers from each cooperative).

Table 3. Characteristics of geographic regions in which the cooperatives examined are situated.

County	Cooperative	NCRPPH	Altitude (m)	Mean Annual Rainfall (mm)	Soil Type Quality	Crop Production System Zone	Number of Members	Sub-Frigid Zone (°C)	Growth Period (days)
Bei’an	Bei’an 1T	Yes	320–750	566	Chernozem, Black soil, meadow soil and others	Soybean, corn, rice,	551	−4–8	130–150
	Bei’an 2T	Yes					509
	Bei’an 1C	No					416
	Bei’an 2C	No					329
	Bei’an 3C	No					547
Wudalianchi	Wudalianchi 3T	Yes	270–510	516	Chernozem, Black soil, meadow soil	Soybean, corn, rice, durum wheat	638	−5–8	130–150
	Wudalianchi 4T	Yes					963
	Wudalianchi 4C	No					1168
	Wudalianchi 5C	No					649
	Wudalianchi 6C	Yes					785
Nenjiang	Nenjiang 5T	Yes	200–500	496	Chernozem, Black soil, meadow soil	Soybean, corn, rice, durum wheat	494	−4–8	130–150
	Nenjiang 6T	Yes					518
	Nenjiang 7C	No					653
	Nenjiang 8T	Yes					684
Xunke	Xunke 9T	Yes	100–110	650	Chernozem, Black soil	Soybean, corn, rice, durum wheat	673	−5–7	130–150
	Xunke 10T	Yes					486
Sunwu	Sunwu 7T	Yes	110–755	541	Chernozem, Black soil, meadow soil	Soybean, corn, rice, durum wheat	212	−4–8	130–150
	Sunwu 8C	No					285
	Sunwu 9C	No					373

Note: All cooperatives are part of the Nenjiang River Basin. The landform is mainly plain and the climate is mainly cold temperate continental monsoon climate. Source: authors.

describes the basic situation of the selected cooperatives, and Figure 3 reports the location of treated and control cooperatives.

We used the following items to identify the treated group: (i) they joined an NCRPPH cooperative; (ii) took part in at least one NCRPPH activity; (iii) agreed to crop rotation. There is quite a high coincidence in the treated group, as shown in

Table 4. Differences and overlaps according to different processes.

No Less Than One Piece of Land Shall Be Used for Crop Rotation	Join at Least One NCRPPH Event
	Total	Treated	Non-Treated
Total	404 (55.12%)	329 (44.88%)	404 (55.12%)
Treated	310 (42.29%)	292 (39.84%)	18 (2.46%)
Non-treated	386 (52.66%)	37 (5.05%)	386 (52.66%)

Source: authors.

.

4.2. Outcome Variables

Several modules in the survey questionnaire relate to agricultural activities and socioeconomic outcomes. From 2015 (just before the program started) to 2019, we gathered data on inputs, outputs, assets, and social capital.

To assess the impact of the NCRPPH on agricultural operations, three outcome variables were chosen (Table 4). The first is growth in cereal production per hectare between 2015 and 2019 (using 2015 constant prices)—this result is used to assess the project’s success in terms of output. The second result is the net value of crop production per hectare per family labor unit (FLU), which is used as a proxy for household income. The net value is calculated as the gross value minus the non-labor input cost and the hired labor cost. The family labor group (FLG) was determined by adjusting for the work of family members engaging in agricultural activities. All interviewed farmers are mainly engaged in agricultural production. The descriptive statistics in

Table 5. Outcome variables—descriptive statistics (2019).

Variables	Total		Treated		Control
	Mean	s.d.	Mean	s.d.	Mean	s.d.
Production-related outcomes
Crop production with a net value (10,000 RMB/ha/FLU)	0.38	1.29	0.57	1.44	0.27	1.06
Percentage of cereal production sold through cooperative	0.11	0.19	0.23	0.18	0.0	0.18
Growth in cereal production (2015–2019) (10,000 RMB/ha)	0.19	1.25	0.39	1.24	0.09	1.12
Food Security-Related Outcomes
Dietary diversity score for HH	9.51	1.74	9.72	1.76	9.44	1.72
HH food insecurity access scale (HFIAS)	0.13	0.52	0.04	0.24	0.18	0.65
“Eating fewer meals” occurs frequently(often sometimes never)	0.04	0.24	0.01	0.10	0.07	0.28
“Eating smaller meals” occurs frequently (often sometimes never)	0.06	0.36	0.01	0.12	0.14	0.35
Education-Related Outcomes
Percentage of girls in HH aged 6–18 in education	0.89	0.03	0.93	0.02	0.87	0.02
Percentage of boys in HH s aged 6–18 in education	0.85	0.01	0.86	0.03	0.83	0.01
Percentage of children in HH aged 6–18 in education	0.88	0.01	0.90	0.01	0.86	0.01
Land Use-Related Outcomes
Crop rotation practice = 1	0.57	0.50	0.55	0.50	0.56	0.50
Other cereals’ proportion of total cereal area	0.17	0.18	0.11	0.12	0.18	0.20
Soybeans’ proportion of total cereal area	0.59	0.33	0.38	0.29	0.67	0.28
Corn’s proportion of total cereal area	0.92	0.13	0.90	0.11	0.88	0.14

Note: RMB stands for the people’s Republic of China (the national currency). RMB 10,000 is approximately EUR 1224.3 based on 2015 exchange rate. Source: authors.

also support this situation: 80–82% of the land was used for planting cereals before attending NCRPPH. To illustrate agriculture’s proportional importance, the average annual net household income from cereal production is RMB 22,000.

The third variable is the proportion of cereal production value sold via cooperatives. To gain a more homogeneous unit of measurement, the proportion of cereal value should be chosen. All empirical studies were also applied to the quantity of production as a robustness check, yielding remarkably comparable results. This result enables the identification of the NCRPPH’s efficacy in creating a viable alternative to local markets and middlemen/intermediaries. The legal effect of the agricultural contract makes farmers face significant monetary loss when selecting other sales channels; therefore, they sell almost all their products to the designated company. This argument is supported by observations made during the qualitative interviews.

Furthermore, this paper investigates the impact of the NCRPPH on the following two critical welfare outcomes: food security and educational chances. Two separate factors are used to assess food security. We first consider the family’s dietary variety score, which is calculated as the number of various food groups consumed the day before the interview [60]—which may range from 0 (no food group consumed) to 16 (all food groups consumed). The next indicator is calculated based on the Household Food Insecure Access Scale (HFIAS) [61]. This indicator is produced by combining self-reported data on how frequently people utilize coping methods, which range from moderate (eating one fewer meal) to excessive (experiencing any type of food deficiency). Since roughly 96% of the households surveyed have never used any of these coping techniques, there are no extreme cases of food insecurity. We also investigated how the project affected how often farmers used two of the coping techniques (the other two remained unchanged); the values of these variables are 0, 1, or 2, depending on whether the farmers have never, occasionally, or frequently experienced the incident, respectively. The chosen outcome in terms of education is the percentage of family members aged 6–18 who are currently enrolled in some sort of formal education. This variable is also broken down by gender.

Finally, this paper considers the potential crowding-in impacts of crop rotation expansion. Heilongjiang Province must secure national food security. That is, it will progressively expand to increase the space available for crop rotation without interfering with other uses. Three possible substitution effects were investigated. First, crop rotation area expansion could have resulted in an increase in crop types, as well as a shift in crop rotation from grazing/fallow to cereals and cereals to pulses (i.e., a practice to maintain and restore soil fertility). Second, horticulture and other non-cereal crops could have been crowded in. Finally, corn and soybeans or other minor cereals may have been cycled.

4.3. Econometric Strategy

When specific players (e.g., farmers) are free to embrace (or refuse) an innovation, the application of propensity score-matching approaches could lead to a skewed estimation of the impact, due to the selection of unobservable factors (e.g., risk awareness, entrepreneurship, self-confidence, etc.) [15]. An instrumental variable (IV) approach has often been used as a remedy to assess the impact of interventions encompassing smallholders’ involvement in agro-industrial activities or their adoption of improved crop rotation varieties [62,63].

Considering this, it is worth delving into the rationale for implementing an IV method to examine the impact of the NCRPPH. The following is the setting:

$$Y=D\beta +X\gamma +\epsilon$$

(1)

$$D=Z\pi +\nu$$

(2)

where Y is the outcome variable, D is the treatment variable (which is associated with other unobservable variables and hence endogenous), and X is a matrix of additional covariates; Z denotes the instrument(s) that are connected to the endogenous therapy but not to the result Y directly. Membership in an NCRPPH cooperative is a suitable candidate for an IV in this context. Given the instrument’s 66.7% compliance rate, the instrument’s strength should not be an issue. Members of NCRPPH cooperatives were all exposed to the prospect of joining the therapy in this study, while members of non-NCRPPH cooperatives were not. For each town, there is a single cooperative. As a result, the assignment of the intention to treat might be regarded as random from an individual’s perspective. However, in the pre-project period, there are statistically significant differences between farmers who were members of NCRPPH cooperatives and those who were members of non-NCRPPH cooperatives, as seen in the first three columns of Table 5 (2015). On average, the former had somewhat more land and productive assets, as seen by greater pre-NCRPPH soil production, and were less vulnerable to natural risks such as floods and frost. This suggests that the instrument might have an impact on outcomes via pathways other than NCRPPH participation, thereby breaching the exclusion limitation assumption.

The instrument is unlikely to be exogenous in this context. As a result, we used three alternative ways to loosen the constraint assumption. First, we carried out a sensitivity analysis using the method described by Conley et al. [64]. Briefly, the impact of violating the limitation assumption may be modeled by defining multiple ranges of values for the coefficient (i.e., specifying the maximum and minimum values). We may generate confidence intervals for the impact coefficient in this manner.

$$Y=D\beta +X\gamma +Z\vartheta +\epsilon$$

(3)

$$D=Z\pi +\nu$$

(4)

The second technique entails a two-step process. First, we utilized pre-treatment variables to perform a PSM estimate on the intention to treat (i.e., Z in Equation (1)) as a treatment variable using a caliper (neighbor matching without replacement). This initial stage is intended to control pre-project imbalances in important variables. A caliper of 0.025 was utilized to calibrate the matching, resulting in a more than adequate decrease in the pre-matching bias.

Table 6. Differences before and after matching between farmers in NCRPPH and non-NCRPPH cooperatives.

Variable (pre-NCRPPH)	Pre-Matching Means			Post-Matching Means
	(a) NCRPPH Coop	(b) Non-NCRPPH Coop	(a)–(b)	(c) NCRPPH Coop	(d) Non-NCRPPH Coop	(c)–(d)
Household characteristics
Age of a householder	41.71	44.11	−2.4 *	42.91	42.81	0.1
Education years of a householder	5.55	5.34	0.21	5.47	5.44	0.03
Adult amount	5.86	5.32	0.54 **	5.53	5.62	−0.09
Householder is a man = 1	0.92	0.95	−0.03	0.96	0.95	0.01
Householder is an ethnic minority = 1	0.63	0.46	0.17 **	0.57	0.59	−0.02
Social involvement
Social involvement with a score of 0–7	1.97	1.83	0.14 *	1.82	1.82	0
Trust in cooperatives with a score of 1–4	2.92	2.94	−0.02	2.95	2.98	−0.03
Involvement in cooperatives with a score of 1–4	3.08	2.98	0.1	2.91	2.93	−0.02
Agroecological characteristics
Land share for growing cereals	0.88	0.87	0.01	0.88	0.84	0.04
The farm has experienced flood fighting = 1	0.37	0.42	−0.05 ***	0.42	0.41	0.01
The farm experienced frost = 1	0.06	0.13	−0.07 **	0.15	0.14	0.01
The farm experienced a drought = 1	0.09	0.04	0.05	0.06	0.05	0.01
Crop rotation area yields (ha)	28.81	26.45	2.36 *	25.64	25.39	0.25
Infrastructure
Time to a health center	24.65	21.55	3.1 *	22.03	22.38	−0.35
Time to the Cooperative Office	13.59	14.9	−1.31 *	14.24	13.96	0.28
Get the time to the mobile network	3.44	4.2	−0.76	4.09	3.97	0.12
Household asset
Farm size (ha)	3.96	3.54	0.42 **	3.77	3.62	0.15
Value of HH productive plants (RMB)	1557	1018	539 **	1049	1225	−175
Value of agricultural assets (RMB)	888	834	54	977	926	51 *
Value of owned livestock (RMB)	17361	17620	−259	17211	16829	382
HH has access to piped water = 1	0.41	0.35	0.06 *	0.4	0.4	0
HH owned No less than 1 PC = 1	0.43	0.42	0.01	0.45	0.43	0.02
HH owned No less than 1 mobile = 1	0.31	0.32	−0.01	0.34	0.32	0.02
HH owned No less than 1 heating equipment = 1	0.74	0.61	0.13 ***	0.63	0.69	−0.06
HH owned No less than 1 vehicle = 1	0.12	0.16	−0.04	0.16	0.13	0.03

***, **, * are significant at 1%, 5%, and 10% level, respectively.

describes the variables utilized for matching and clearly illustrates that post-matching mean differences are not significant, and that the matching technique was able to re-balance pre-matching systematic discrepancies between project and control regions. This is also supported by the typical matching quality statistics (

Table 7. Matching quality statistics.

Sample	Pseudo R2	LR chi2	Mean Stand. Bias	Median Stand. Bias	% Variance	Rubin’s R	Rubin’s B
After matching	0.02	6.13	3.4	3.2	22.4	1.17	21
Before matching	0.11	101.87	12.3	10.6	77.1	0.96	34

Source: authors.

). The critical values proposed by Rubin are: post-matching, Rubin’s B is below 25, and Rubin’s R is within the range 0.5–2 [65]. Rubin’s B is the absolute standardized difference in the means of the linear index of the propensity score in the treated and (matched) non-treated group; Rubin’s R is the ratio of treated to (matched) non-treated variances of the propensity score index.

After eliminating non-matched observations, the resultant sub-sample consists of the following 603 units: 161 treatment and 442 control (including 90 non-compliers living in NCRPPH cooperatives). The IV estimate was then performed on this sub-sample.

Frölich [66] suggested a totally non-parametric estimator for estimating the local average treatment effect (LATE) with covariates as a third technique for dealing with a powerful but possibly endogenous instrument—only a subset of variables X are expected to fulfill the exclusion criterion.

Leto β be the LATE estimate. The conditional mean function is as follows:

$$\mathrm{mz}\left(\mathrm{x}\right)=\mathrm{E}[\mathrm{Y}|\mathrm{X}=\mathrm{x};\mathrm{Z}=\mathrm{z}]$$

(5)

$$\mathsf{\mu }\mathrm{z}\left(\mathrm{x}\right)=\mathrm{E}[\mathrm{D}|\mathrm{X}=\mathrm{x};\mathrm{Z}=\mathrm{z}]$$

(6)

The ratio between two matching indicators (βis) as follows:

$$\overline{\beta }=\frac{{{\displaystyle \sum }}_i^{}{m}_1\left(X_i\right)-m_0\left(X_i\right)}{{{\displaystyle \sum }}_i^{}{\mu }_1\left(X_i\right)-{\mu }_0\left(X_i\right)}$$

(7)

We are certain that by using these multiple estimating methodologies, we can eliminate a large portion of the possible biases. However, given the use of retrospective data and a less-than-ideal identification approach, we cannot rule out the possibility that the project’s impacts results are skewed upwards.

5. Results and Discussion

Table 8. Estimates of the impact of NCRPPH—sensitivity analysis.

ϑmin	ϑmax	Growth in Cereal Production			Percentage of Cereal Production Sold through Co-Op			Crop Production with a Net Value
		β	β 95% Confidence Interval		β	β 95% Confidence Interval		β	β 95% Confidence Interval
			β −1.96se	β +1.96se		β −1.96se	β +1.96se		β −1.96se	β +1.96se
0	0	0.42	0.17	0.67	0.17	0.14	0.20	0.29	0.21	0.37
0	0.1	0.35	0.18	0.52	0.09	0.05	0.13	0.21	0.15	0.27
0	0.2	0.32	0.23	0.41	0.02	−0.09	0.13	0.14	0.11	0.17 ●
0	0.3	0.30	0.29	0.31	−0.06	−0.25	0.13	0.06	0.05	0.07 ●
0	0.4	0.27	0.21	0.33	−0.13	−0.39	0.13	−0.01	−0.03	0.01 ●
0	0.5	0.25	0.12	0.38	−0.20	−0.53	0.13 ●	−0.08	−0.12	−0.04 ●
0	0.6	0.23	0.02	0.44	−0.27	−0.67	0.13 ●	−0.15	−0.21	−0.09 ●
0	0.7	0.20	−0.09	0.49 ●	−0.34	−0.81	0.13 ●	−0.22	−0.31	−0.13 ●
0	0.8	0.18	−0.17	0.53 ●	−0.41	−0.95	0.13 ●	−0.29	−0.40	−0.18 ●
0	0.9	0.14	−0.33	0.61 ●	−0.49	−1.11	0.13 ●	−0.37	−0.51	−0.23 ●

● = 0 is included in the 95% confidence interval. Source: authors.

shows the sensitivity analysis and standard IV estimations. Values are displayed for the three outcomes for various values (of values with = 0), which is the conventional IV impact estimate—shown in the first row of the table. These preliminary findings indicate that treated farmers saw an RMB 4232 increase in the value of their gross cereal yield per hectare. Furthermore, the project enhanced the net value of production per FLG by approximately RMB 2945 per hectare. These three indicators confirm the positive impact of the NCRPPH on production, but whether the results are robust requires verification in subsequent studies.

Lastly, in the treated group, the share of production sold through the cooperative increased by 20% (from an average of 2% in the control group). As shown in Table 6, families in project areas are often wealthier, more endowed with land, and less vulnerable to climatic threats than those in control regions. This justifies the hypothesis that ϑ ≥ 0 –supports the hypothesis that living in an NCRPPH area is positively linked to better health (actual participation in the NCRPPH activities is not included).

The sensitivity analysis shows that the impact estimate is still significantly different from zero with quite high values of ϑ. For example, impact estimates regarding production growth “tolerate” a ϑmax = 0.6 starting from a given baseline β = 0.42 with ϑmin = 0.

Secondly, we estimated IVs using only data matching the closest neighbor method on a subsample of treated and control farmers. The IV estimate findings for the subsample (see

Table 9. Estimates of the impact of NCRPPH with PSM + IV strategy.

Outcome	β	s.e.	p-Value	A-R test	p-Value	F-Stat	p-Value
Growth in cereal production	0.38	0.02	0.000 ***	80.61	0.000	8.93	0.000
Crop production with a net value	0.30	0.03	0.000 ***	39.79	0.000	6.87	0.000
Percentage of cereal production sold through co-op	0.17	0.01	0.000 ***	75.74	0.000	4.63	0.000

*** is significant at the 1% level, respectively. Source: authors.

) are consistent with those obtained using the usual IV technique.

Finally,

Table 10. Non-parametric LATE estimates of the impact of NCRPPH.

Outcome	β	s.e.	p-Value
Growth in cereal production	0.381	0.0178	0.000 ***
Crop production with a net value	0.297	0.0346	0.000 ***
Percentage of cereal production sold through co-op	0.174	0.0318	0.000 ***

*** is significant at the 1% level, respectively. Source: authors.

shows the non-parametric LATE estimates computed using the estimator suggested by Frölich [666]. The estimations are consistent with the baseline IV estimates as well as the IV with PSM estimates.

Overall, the results showed that the test was robust and confirmed that the +NCRPPH had a positive effect on all three results. It also implies that the results are maintained with a less restrictive definition of the exclusion limitation assumption.

5.1. Heterogeneity Analysis

To understand how effective a project such as NCRPPH is, we need to find out if some groups benefit more than others. Herrmann discovered that involvement in sugarcane out-grower schemes in Tanzania resulted in higher revenue increases among land-rich farmers than among land-poor farmers [24]. On the other hand, in Kenya, Rao and Qaim discovered that participation in retail channels benefits land-poor farmers disproportionately [22].

This section investigates changes in the impact of NCRPPH based on the following farm size: 0–2 ha, 2–4 ha, and greater than 4 ha (

Table 11. Farm size groups and treatment status.

Farm Size	Total	Treated	Control
0–2 ha	213	45	168
2–4 ha	275	97	178
4+ ha	223	104	119

Source: authors.

). We selected 0–2 ha category because The Food and Agriculture Organization Corporate Statistical Database and other scholars use the same classification [67,68]. The 2–4 ha and 4+ ha categories are determined to have categories of comparable size/grade, which cause the results to have no significant difference from Table 10. Farm size is used to gauge the economic performance and productivity of a household. The IV with PSM and LATE approaches were used for this heterogeneity analysis. As indicated in

Table 12. Estimates according to farm size.

Outcome	Farm Size (ha)	PSM + IV			LATE
		β	s.e.	p-Value	β	s.e.	p-Value
Growth in cereal production	(a) 0–2	0.48	0.11	0.000 ***	1.46	0.27	0.000 ***
	(b) 2–4	0.37	0.05	0.000 ***	1.17	0.23	0.000 ***
	(c) 4+	0.39	0.07	0.000 ***	1.19	0.19	0.000 ***
Crop production with a net value	(a) 0–2	0.33	0.07	0.000 ***	1.08	0.17	0.000 ***
	(b) 2–4	0.27	0.08	0.000 ***	0.70	0.21	0.000 ***
	(c) 4+	0.26	0.07	0.000 ***	1.02	0.22	0.000 ***
Percentage of cereal production sold through co-op	(a) 0–2	0.26	0.08	0.000 ***	0.19	0.07	0.000 ***
	(b) 2–4	0.12	0.05	0.000 ***	0.14	0.05	0.000 ***
	(c) 4+	0.21	0.04	0.000 ***	0.21	0.06	0.000 ***

*** is significant at the 1% level, respectively. Source: authors.

, the NCRPPH has a positive and statistically significant effect across all three groups and results. As indicated in

Table 13. Tests of differences in project’s impact across farm size categories with PSM + IV.

Outcome	Difference	βi–βj	s.e.	p-Value
Growth in cereal production	Group(a)–group(b)	0.35	0.41	0.386
	Group(a)–group(c)	0.34	0.43	0.369
	Group(b)–group(c)	−0.01	0.28	0.943
Crop production with a net value	Group(a)–group(b)	0.22	0.28	0.472
	Group(a)–group(c)	0.24	0.24	0.335
	Group(b)–group(c)	0.02	0.31	0.907
Percentage of cereal production sold through co-op	Group(a)–group(b)	0.11	0.07	0.183
	Group(a)–group(c)	0.06	0.07	0.521
	Group(b)–group(c)	−0.05	0.06	0.262

Source: authors.

, the project did not have a large impact on differently sized farms. As a result, contrary to the findings of Herrmann, Rao, and Qaim [22,24], we conclude that the project benefits both land-rich and land-poor farmers equally. We suspect that this is because the NCRPPH provides participants with similar distribution channels and planting technical guidance, thus avoiding the disadvantages of small farmers in sales and production.

5.2. Analysis of the NCRPPH’s Effects on Land-Use Outcomes

Table 14. PSM + IV estimates of impact on land-use outcomes.

Outcome	β	s.e.	p-Value
Crop rotation practice = 1	0.04	0.07	0.628 **
Cereals’ proportion of total cereal area	0.05	0.06	0.152
Soybeans’ proportion of total cereal area	0.71	0.18	0.000 ***
Othres’ proportion of total cereal area	0.19	0.04	0.000 ***

*** and ** are significant at the 1% and 5% level, respectively. Source: authors.

shows that cereal crops are the most commonly substituted for other crops, while crop rotation practices have largely continued to change. However, participating in the NCRPPH has increased land used for soybeans and other cereals (to a lesser extent). In other words, soybeans are crowding out other crops, such as corn. Grazing, horticulture, and fallow fields were not affected as much. This is different from empirical results obtained by Peltonen-Sainio in Finland [69]. The reason may be the different planting and breeding structures. Heihe is currently in the stage of expanding animal husbandry. Affected by policy subsidies, farmers are less likely to reduce grazing. On this basis, we conducted a heterogeneity analysis (

Table 15. Estimates of the impact of the NCRPPH on land-use outcomes, according to farm size.

Outcome	Farm Size (ha)	β	s.e.	p-Value
Crop rotation practice = 1	(a) 0–2	−0.154	0.137	0.275
	(b) 2–4	−0.038	0.101	0.735
	(c) 4+	0.092	0.085	0.355
Cereals’ proportion of total cereal area	(a) 0–2	0.103	0.036	0.011 **
	(b) 2–4	0.104	0.017	0.000 ***
	(c) 4+	0.012	0.015	0.506
Soybeans’ proportion of total cereal area	(a) 0–2	−0.253	0.076	0.000 ***
	(b) 2–4	−0.264	0.052	0.000 ***
	(c) 4+	−0.244	0.042	0.000 ***
Others’ proportion of total cereal area	(a) 0–2	−0.207	0.102	0.053 *
	(b) 2–4	−0.135	0.081	0.048 **
	(c) 4+	−0.182	0.094	0.043 **

***, **, * are significant at the 1%, 5%, and 10% level, respectively. Source: authors.

). The NCRPPH has no impact on how likely it is to practice crop rotation, how many soybeans make up total land, or how many other cereals make up the total cereal area. Moreover, there is a difference in the proportion of cereals to total agricultural land, as follows: the NCRPPH’s effects are stronger for farmers with 0–2 ha and 2–4 ha of land. This explains why small farmers who want to grow soybeans should reduce the production of other grains.

5.3. Analysis of the NCRPPH’s Effects on Food and Education

Next, there are some hidden welfare effects worth considering, such as food safety and education.

In

Table 16. Estimation of PSM + IV impact on food security and education.

Outcome	β	s.e.	p-Value
Education
Percentage of children in HH aged 6–18 in education	0.08	0.02	0.089 *
Percentage of boys in HH aged 6–18 in education	0.07	0.06	0.076 *
Percentage of girls in HH aged 6–18 in education	0.12	0.05	0.023 **
Food Security
HFIAS	−2.38	1.19	0.049 *
“Eating smaller meals” occurs frequently (often/sometimes/never)	−1.39	0.68	0.046 **
“Eating fewer meals” occurs frequently(often/sometimes/never)	−0.75	0.93	0.432
Dietary diversity score for HH	−0.11	0.13	0.518

** and * are significant at the 5% and 10% level, respectively. Source: authors.

, we use the enrollment proportion of family members aged 6–18 to measure the impact of NCRPPH on education. The result is positive, and the impact is stronger among girls. In particular, the sample included 286 control observations and 223 treatment observations for female samples, 243 control observations and 269 treatment observations for male samples, and finally, 312 control observations and 251 treatment observations for the combined group.

Even though the coefficient is significant only at the 10% level, treated households scored lower on the household food insecurity access scale. Households in the treated group ate smaller meals less often than those in the control group (coefficient of 5%). Additionally, they use fewer coping strategies, “eating fewer meals” than control households; the difference, however, is not statistically significant. Finally, there was no statistical significance in the investigation of food types.

The heterogeneity analysis based on the farm scale is shown in

Table 17. Estimates of the impact of the NCRPPH on well-being outcomes, according to farm size.

Outcome	Farm Size (ha)	β	s.e.	p-Value
Education
Percentage of children in HH s aged 6–18 in education	(a) 0–2	0.069	0.091	0.445
	(b) 2–4	0.044	0.062	0.484
	(c) 4+	0.084	0.053	0.117
Percentage of boys in HH aged 6–18 in education	(a) 0–2	0.137	0.099	0.168
	(b) 2–4	−0.095	0.076	0.216
	(c) 4+	0.003	0.072	0.586
Percentage of girls in HH aged 6–18 in education	(a) 0–2	0.061	0.117	0.601
	(b) 2–4	0.244	0.091	0.008 ***
	(c) 4+	0.004	0.043	0.814
Food Security
HFIAS	(a) 0–2	−0.126	0.211	0.547
	(b) 2–4	−0.036	0.106	0.732
	(c) 4+	−0.141	0.078	0.073 *
“Eating smaller meals” occurs frequently (often sometimes never)	(a) 0–2	−0.085	0.113	0.446
	(b) 2–4	−0.025	0.061	0.674
	(c) 4+	−0.111	0.053	0.038 **
“Eating fewer meals” occurs frequently(often sometimes never)	(a) 0–2	−0.042	0.087	0.633
	(b) 2–4	0.004	0.052	0.936
	(c) 4+	−0.035	0.036	0.333
Dietary diversity score for HH	(a) 0–2	−0.663	0.593	0.263
	(b) 2–4	−0.476	0.335	0.156
	(c) 4+	0.432	0.313	0.168

***, **, * are significant at the 1%, 5%, and 10% level, respectively. Source: authors.

. The project has no significant impact on children’s education as a whole. On the other hand, among girls, the impact of the NCRPPH is obviously concentrated in the category with a farm area of 2–4 ha. In terms of food safety, the farm area did not cause the NCRPPH project to have a differential impact on the family dietary diversity and frequency of eating fewer meals scores. On the contrary, the impact on the other two food security outcomes was triggered by farmers in the largest farm scale category. This is at odds with Jones’ findings that participation in crop rotation by small farmers leads to richer and more nutritious diets [70].

6. Concluding Remarks and Policy Implications

In recent years, China’s grain output and agriculture have been steadily increasing and developing, respectively, and have made an important contribution to collective economic and social development. However, China’s agricultural development model is relatively extensive, and a series of problems such as overexploitation of agricultural resources, overuse of agricultural investment, overexploitation of groundwater, and superposition of agricultural internal and external source pollution are becoming increasingly prominent.

The sustainable development of agriculture faces major challenges. However, these can be solved through accelerating the transformation of agricultural development mode, promoting ecological restoration and governance, and promoting the sustainable development of agriculture. Therefore, taking advantage of the abundant food supply in the domestic and foreign markets at the present stage, the implementation of crop rotation in some areas is not only conducive to the recuperation of cultivated land and the sustainable development of agriculture but also conducive to balancing the contradiction between food security and agriculture-led industrialization and the well-being of farmers.

Using large-scale household survey data from 2019, this paper analyzed the impact of the NCRPPH project on grain production and sales channels. In addition, we also analyzed the impact of the project on food safety, education welfare, and land use. In terms of experimental methods, we selected three widely used estimation strategies that were implemented to ensure the credibility of the results.

The three experimental results confirm the positive impact of the NCRPPH project on the gross grain production and net worth per hectare, and the impact on differently sized farms is positive. At the same time, the project was shown to have no negative impact on the rotation, and the increase in soybean rotation area also brings an increase in soybean yield. Other estimates also verify the positive contribution of the NCRPPH project to education and food security. Although the evaluation of the positive effect of the NCRPPH project requires long-term official research, the investigation of the project in this study shows its potential in improving agricultural CC and farmers’ living standards.

However, two points should be clarified before repeating or scaling up this type of intervention. Firstly, while the statistics appear to indicate that the NCRPPH does not help land-rich farmers proportionately, the issue of project participation remains. Our analysis demonstrates that land and asset ownership is critical in determining who participates in the project. While this topic warrants more examination, we believe that large economies of scale are unnecessary for this type of initiative to succeed. Moreover, we believe the NCRPPH’s innovations attracted fewer risk-averse, wealthier farmers in the beginning, but additional research is needed to substantiate this claim.

Second, when adjusting the NCRPPH project, we should fully consider that the farm scale in Heilongjiang Province is higher than the national average level and the soil conditions are more suitable for agricultural production. There is significant differentiation among farmers’ groups in Heilongjiang Province. Different farmers have different interests in crop selection and temporal and spatial layout. Therefore, how to cooperate with diversified farmers’ groups and build a cultivated land rotation system according to local conditions is particularly important. For example, in the Nenjiang Plain of this research institute, according to climate and environment characteristics, the maize soybean rotation system is implemented in the north, and the grain soybean, grain economy, grain feeding, and other rotation systems are established in the south and west.

The cooperation between crop rotation and other conservation tillage methods is a new direction for follow-up research. One decade-long experiment proved that no-tillage and subsowing can effectively promote corn growth and yield [71]. Some scholars have started focusing on the synergistic effect of crop rotation with the application of organic fertilizer, straw returning, no-tillage, and other conservation tillage methods [72,73]. The Chinese government has also started encouraging farmland construction, water and soil conservation, water conservancy projects, and agronomic measures to comprehensively harness land. While this is a trend, it should be fully tested and evaluated before implementation.

In general, the development of small-scale farmers in Heilongjiang Province provides more opportunities. However, it remains a challenge to improve output while ensuring quality and establishing close cooperation with national food producers.