4.1.2. Robustness Test

Simultaneous simulation of "History Replication" and "History Divergent" for the same industry is a commonly used method to test the robustness of the setting of the parameters' values [41,44,62]. The characteristics of the biodiesel industry under the baseline scenario are simulated by adjusting the parameters and setting the initial values according to the stylized facts of the biodiesel industry in China. The simulation results are shown in Figures 5 and 6. The characteristics of the biodiesel industry evolution are significantly different from those of the fuel ethanol industry evolution. This means that the simulation results of the coevolutionary model set in Section 3.2 are indeed affected by the parameter settings. Therefore, the setting of the parameters' values and the initial values under the baseline scenario can be used for further policy analysis.

**Figure 5.** Production change in the biodiesel industry.

**Figure 6.** Change of the technical efficiency of the biodiesel industry.

### *4.2. Policy Impacts Simulation*

### 4.2.1. The Impacts of Entry Regulation

Entry regulation is one of the most important industry regulation policies. The governmen<sup>t</sup> often uses this kind of policy to avoid overcapacity or other negative social or economic impacts due to overheating industrial development. Entry regulation has also been used in the managemen<sup>t</sup> of the fuel ethanol industry. How effective is this policy? Does this policy have any other impacts on the industry's growth while achieving its policy objectives? These questions are essential for evaluating the performance of regulation policy. To shed some light on the above questions, we analyzed the impacts of entry regulation on the fuel ethanol industry's evolution using the simulation method. The simulation results are shown in Figures 7–9.

**Figure 7.** Impact of entry regulation on the demand for feedstock.

**Figure 8.** Impact of entry regulation on the price of feedstock.

**Figure 9.** Impacts of entry regulation on fuel ethanol production.

In order to prevent the sharp rise in grain prices due to the rapid development of the traditional grain-based fuel ethanol industry, the Chinese governmen<sup>t</sup> has implemented an entry regulation that restricts the new establishment of grain-based fuel ethanol firms. The results in Figure 7 show that this regulation policy effectively suppresses the demand for traditional feedstock (e.g., corn and wheat). However, an entry regulation has no significant impact on grain price (shown in Figure 8), because the demand for grain as feedstock for the fuel ethanol industry is only a small part of the total grain demand. Figure 9 shows that the growth of the fuel ethanol production decreases in the short term due to entry regulations. However, in the long run, the restraint effect of entry regulation on production will be eliminated due to the creation of new technology firms. Therefore, entry regulation is conducive to

restricting the expansion of the grain-based fuel ethanol industry in the short term, while the long-term impact on the fuel ethanol industry is not too large. Moreover, this policy is helpful in promoting new technology and accelerating technology transformation.

### 4.2.2. The Impacts of Production Subsidy

A production subsidy is one of the most common policies for the governmen<sup>t</sup> to promote the development of an emerging industry. Whatever the form of the subsidy, the mechanism of the subsidy is to avoid the losses caused by an immature technology and market in the early stages of the industry so that incumbent firms can continue to produce and improve their technology. With the improvement of technology and the market environment, firms will face competition in the market and make normal profits. Figures 10 and 11 show the impacts of a production subsidy on the output and the number of firms in the fuel ethanol industry, respectively.

**Figure 10.** Impact of the production subsidy on the output.

**Figure 11.** Impact of the production subsidy on the number of firms in the industry.

As shown in Figures 10 and 11, under no production subsidy scenario, firms will continue to enter randomly, and some will withdraw because of losses. Therefore, the industry output and the number of firms will remain at a low level for a long time. However, under the production subsidy scenario, the number of firms entering production will continue to increase, and the output will also continue to rise. The figures also show that the growth rate of the output declined significantly because of the entry regulation on grain-based fuel ethanol firms after the 20th period. Only when new technology firms enter the industry does the output increase significantly again. Therefore, although the simulation results indicate that a subsidy promotes industry growth, this is not a general conclusion. A production subsidy only promotes the growth of subsidized firms and the industries constituted by these firms. When there are different technological routes in the development of the fuel ethanol industry, the implementation of a single subsidy policy may not promote—or even hinder—the development of the industry.

### 4.2.3. The Impacts of R&D Subsidy

A subsidy for R&D activities helps to accelerate technological progress by increasing R&D investment. Therefore, an R&D subsidy policy is widely used to promote industry growth. In terms of the fuel ethanol industry, there are not only traditional technology firms but also new technology firms and R&D firms in the industry. This section will analyze the differences in the policy effects of R&D subsidy given to different types of firms. To reduce the influences of random factors, the impacts of the R&D subsidy on average technical efficiency industry output were all simulated 10 times. The simulation results are shown in Figures 12–15.

**Figure 12.** Impact of the R&D subsidy for traditional technology firms on average technical efficiency.

**Figure 13.** Impact of the R&D subsidy for R&D firms and new technology firms on average technical efficiency.

The results in Figures 12 and 13 show that the R&D subsidy promotes the technical efficiency of the fuel ethanol industry significantly, regardless of the R&D subsidy for traditional technology firms, R&D firms, or new technology firms. However, in terms of industrial scale, the results in Figures 14 and 15 show that the R&D subsidies have no significant impact on industry output, regardless of the subsidy to traditional technology firms or new technology firms. The reason for this result is that firms in the fuel ethanol industry are all in the red and obtain their production subsidy from the government. Therefore, the improvement of firms' technical efficiency, which is partly caused by the R&D subsidy, has little influence on the output decisions of firms. This means that when firms are in the red, the influence mechanism of the R&D subsidy to promote a firm's output will be hindered.

**Figure 14.** Impact of the R&D subsidy for traditional technology firms on industry output.

**Figure 15.** Impact of R&D subsidies for new technology firms on industry output.

### 4.2.4. The Impacts of Ethanol Mandate

In order to accelerate the development of the fuel ethanol industry in its early stages, the Chinese governmen<sup>t</sup> implemented an ethanol mandate policy that required the gasoline sold in pilot cities to contain 10% fuel ethanol. Will this policy help the development of the fuel ethanol industry? To answer this question, we simulated the impacts of the ethanol mandate policy on the output of the fuel ethanol industry. The simulation results are shown in Figure 16.

**Figure 16.** Impact of the ethanol mandate on the output of the fuel ethanol industry.

The result in Figure 16 shows that when there is no production subsidy for losses of firms, the ethanol mandate has no obvious impact on industry output. However, under the subsidy loss scenario, the ethanol mandate will obviously accelerate the development of the fuel ethanol industry. The reason for this phenomenon is that fuel ethanol firms are usually in the red while the industry remains in its embryonic stage. Firms enter the industry randomly and then exit due to continuous losses, so the fuel ethanol industry scale is relatively small. This means that the equilibrium output of the fuel ethanol industry is mainly affected by the number of incumbent firms but not changes in market demand. Therefore, the ethanol mandate has no obvious influence on the output of the fuel ethanol industry, but when firms obtain a loss subsidy, the number of firms will continue to increase, and the industrial output will increase rapidly and reach a larger scale. In this context, the equilibrium output will be mainly affected by market demand. Therefore, an ethanol mandate policy will be helpful to promote the expansion of the fuel ethanol industry.

### **5. Conclusions and Policy Implications**

The interaction among the fuel ethanol industry, the technology system, and the market system has a substantial effect on the growth of the fuel ethanol industry which plays a key role in the formation of a sustainable energy system in China. However, we know little about the relationships among them and it is difficult to explore the nexus using an econometric method due to the lack of statistics on China's fuel ethanol industry. In order to investigate the coevolutionary relationships between the fuel ethanol industry system, technology system, and market system in China, this paper developed a history-friendly simulation model. The setting of the initial values for the simulation model is based on the historical values of the fuel ethanol industry system, technology system, and market system in China. The parameter values of the model were acquired by adjusting the parameters' values continuously until the simulation results could reflect the stylized facts of the fuel ethanol industry. According to the baseline model, this paper further assessed the impacts of entry regulation, production subsidy, R&D subsidy, and ethanol mandate on the growth of the fuel ethanol industry. The results show that multidirectional causalities reflected by the coevolutionary model developed in this paper can describe the relationships between the fuel ethanol industry, technology system, and market system appropriately. This means that the evolution of the fuel ethanol industry interacts with the evolution of the technology system and the market system. Meanwhile, the evolution of the technology system also interacts with the evolution of the market system. Entry regulation is conducive to weakening the negative economic impacts (e.g., rising grain prices and grain shortages) of the expansion of the grain-based fuel ethanol industry without affecting the long-term total output of the industry. A production subsidy for traditional technology firms is helpful to the expansion of the fuel ethanol industry. However, this subsidy also impedes technology transfer in the industry. In terms of R&D subsidy policy, only when the firms inside the industry are not in the red can an R&D subsidy promote technological progress and further accelerate the growth of the fuel ethanol industry. The ethanol mandate policy has a significant impact on industrial expansion only when a production subsidy policy is implemented at the same time. This policy can also speed up the improvement of new technology efficiency by advancing the creation of R&D firms and new technology firms.

According to the above conclusions, the policy suggestions are as follows: firstly, an assessment of the effects of an exogenous factor on any one of these three systems must consider the cumulative effects caused by the coevolutionary mechanisms. Secondly, in the context of this uncertain industry's economic and social impact in China, implementing an entry regulation would be helpful to promote the steady growth of the fuel ethanol industry. Thirdly, when the optimal technology path has not been determined in the Chinese fuel ethanol industry, it will be necessary to implement a production subsidy combined with a new technology promotion policy to avoid technology lock-in. Fourthly, considering that most firms in the fuel ethanol industry in China are still running under a deficit, an R&D subsidy for production firms needs to be implemented along with a production subsidy for unprofitable firms. Lastly, the Chinese governmen<sup>t</sup> can accelerate the development of the fuel ethanol

industry with the help of an ethanol mandate policy. Additionally, the ethanol mandate policy should be implemented together with the production subsidy policy if a firm is running under a deficit.

### **6. Limitation and Future Research**

One limitation of this study is that it does not check the robustness of the simulation results using a statistical approach. It is di fficult to estimate the values of the parameters in our model using a statistical method for the lack of enough observations and the complex nonlinearity of the history-friendly model. Alternatively, we employed a process used in most literature about the history-friendly model to determine the values of the parameters in our model. The aim of the process is not the specification of the model parameters as close as possible to their actual values, nor is to explain the quantitative values observed in the historical episode under investigation. Rather, the objective is just to seek a set of parameters' values which can generate simulation results qualitatively capturing the stylized facts of the industrial history, because the purpose of history-friendly modelling is to explore the causal relationships and the mechanism between variables in the model [44]. This means there may be many sets of parameters' values that satisfy the requirement. Therefore, after determining a set of parameters' values for the base case, it is necessary to check the robustness of the simulation results. There are two commonly used approaches to test the robustness in the existing literature [41,47]. One is the inspection of individual runs of the model and the analysis of sensitivity to specific parameter values. The other is the running of a history divergent simulation. These two approaches are also used in this study to check the robustness of the results. However, we sugges<sup>t</sup> there should be a more intense discussion of sensitivity analyses using a statistical method in future research because the above two commonly used approaches could not completely reflect the impacts of stochastic components on the simulation results. Some scholars have already made a bit of progress in this direction. For example, Brenner and Murmann initially developed a process to define the values of the parameters that can be observed with precision [61]. Landini et al. introduced a statistical method which can be used to check the robustness of their results with di fferent variation in the ranges of the parameters [46]. However, both statistical methods have their own drawbacks. The first one turns out not to be immediately comprehensible and could not be completely and clearly described due to the usual limits imposed on the number of words for a paper [50]. The second is still arbitrary in the selection of the variation in the range of parameters. Future research can further improve these two statistical methods and then apply them to the testing of the robustness of the simulation results.

**Author Contributions:** Conceptualization, C.B. and J.Z.; methodology, C.B. and W.Z.; software, C.B. and Y.W.; formal analysis, C.B. and J.Z.; data curation, W.Z.; writing—original draft preparation, C.B., J.Z., and W.Z; writing—review and editing, C.B.; visualization, C.B. All authors have read and agreed to the published version of the manuscript.

**Funding:** This research was funded by the fund project of the National Natural Science Foundation of China (Grant No: 71974203), Zhongnan university of economics and law graduate education achievement award cultivation project (Grant No: CGPY201904), interdisciplinary innovation research project (Grant No: 2722019JX002), Zhongnan University of Economics and Law Fundamental Research Funds for the Central Universities (Grant No: 2722020JCT026), the Chinese National Funding of Social Sciences (Grant No: CFA150151), and the Fundamental Research Funds for Shaanxi Normal University (16SZYB34).

**Acknowledgments:** We thank three anonymous referees for very useful comments.

**Conflicts of Interest:** The authors declare no conflict of interest.
