IMPACT OF GOVERNMENT POLICIES ON THE KNOWLEDGE BASE OF SUSTAINABLE SMALL AND MEDIUM-SIZED ENTERPRISES

Small and Medium Enterprises are pivotal in academic research, policy formulation, and economic development because they drive financial growth and create employment opportunities. This study aims to evaluate the factors influencing operational performance in SMEs, focusing on the impact of knowledge and the moderating role of government policies. Utilizing a sample of 200 SME owners in Padang, selected through convenience sampling, data was collected via online questionnaires based on field studies. The analysis, conducted using Structural Equation Modeling-Partial Least Squares, explores the complex relationships between knowledge variables, government policies, and SME operations. Findings reveal that knowledge investment is crucial for the operational success of SMEs, underscoring its importance in business strategy. Government policies also positively influence SME performance, highlighting the need for supportive regulatory environments. This research contributes to a deeper understanding of the factors that support SME success and emphasizes the strategic role of government intervention. The study's novelty lies in its dual focus on knowledge and policy, offering comprehensive insights into their combined effects on SME performance. While the findings are specific to Padang, they suggest broader implications for the role of knowledge and policy in SME development globally. Future research should address the limitations of this study, including its regional focus and variable scope, to enhance our understanding of SME performance on an international scale. This study informs policymakers and SME practitioners and sets the stage for further exploration into the sustainable growth of SMEs worldwide.


INTRODUCTION
Small and Medium Enterprises (SMEs) are now playing an important role in increasing the economic dynamics in the sector of industry.They can significantly develop the economic progress of many countries and maintain long-term and sustainable economic growth (Bendickson et al., 2017) Likewise in terms of job creation and community income as well as local economic development which forces the government to pay attention to the existence of SMEs.(Amoah et al., 2021;Chatterjee, S., & Kar, 2020;Erdin & Ozkaya, 2020;Virglerova, Z., Conte, F., Amoah, J., & Massaro, 2020;Ye, J., & Kulathunga, 2019).Likewise in developing countries, SMEs can contribute to economic growth (Arshad, M., & Arshad, 2019).To encourage this, the presence of the government is also considered to have contributed to establishing various policies aimed at increasing SMEs' entrepreneurship and performance (Jayeola et al. 2022).It can be seen from the results of research carried out by R. Amoah et al. (2022) that the government plays an active role in encouraging SMEs by providing policy by providing a variety of training such as capacity-building training, registration and accounting, safety and health, and environmental management, even though it can not be treated for all SMEs.Similarly, in terms of government technology strategy (Chaudhuri et al., 2023), the government is constantly encouraging SMEs' growth and competitiveness by offering a variety of digital infrastructure and internet connectivity.
Another phenomenon that is opposite to the success of small and medium businesses, is not spared from various factors that hinder it such as financial constraints, management experience, leadership skills, marketing, planning, and education (Akindoju, 2016).According to some scholars, the performance of small and medium enterprises is influenced by several factors, such as financial and non-financial factors (Jebna & Baharudin, 2013;Kemayel, 2015)scale, growth rate, profitability, and industry cohesion of enterprises (Xuan et al., 2020).This is according to the study the research conducted Ebrahim et al., (2023) which declares that lack of financial support, poor management, corruption, lack of training and experience, poor infrastructure, insufficient profits, and low demand for products and services.Interestingly, the presence of the government is considered a factor inhibiting the growth of SMEs (Nkwabi & Mboya, 2019) Likewise in Indonesia, the existence of small and medium enterprises is an important sector of the economy (Twins et al., 2020) because it makes a major contribution to creating jobs and increasing people's income Abduh and Remmang, 2023).Business progress according to R Pramono, LW Sondakh, I Bernardo, (2021) has links and dependencies on sources both capital and education, the need for achievement, and locus of control.However, the performance of small and medium enterprises is also not free from various obstacles, namely the lack of technology in the production sector, human resource capabilities, raw materials, and marketing (Tambunan, 2018).Furthermore (Partala et al. 2024) concluded, that there are internal and external factors that contribute to the success of small and medium enterprises.As well as Hurdawaty & Tukiran, (2024), to enhance MSMEs' competitiveness through digitalization, product and service innovation, business process innovation, and marketing and sales innovation.
Even though there have been more and more research results on the performance of small and medium enterprises, the relationship between knowledge and the performance of SMEs and government policy support has not been conclusive enough, where there are still some previous scholars who found that knowledge variables affect SMEs' performance (Mutairi, Naser and Fayez, 2017;Chaithanapat et al., 2022;Wijaya, Yadewani and Karim, 2022).Meanwhile, other scholars found that knowledge does not affect the performance of small businesses (Ardiana & Brahmayanti, 2010;Echdar, 2014;Putranto and Trihudiyatmanto, 2021).For this reason, the researcher suggests adding government policy variables as a moderator variable because other studies government policy variables have been used as independent variables on the performance of small and medium enterprises (Afrifa and Tauringana, 2015) and also a variable that cannot be controlled by other variables Therefore, the general objective of this study is to analyze the effect of knowledge on the business performance of SMEs in Padang City: Government Policy as a moderating variable.The following are the study's particular objectives: (1) To analyze the impact of knowledge on SMEs' performance (2) To analyze the influence of government policies on the performance of SMEs.(3) To analyze the effect of government policy moderation on the relationship between knowledge and SME performance.

Performance of Small and Medium Enterprises
Typically, performance is measured in terms of the owner's impression of sales growth, profit rate, asset escalation, client base expansion, and so on.As a result, surveys are most commonly used to assess SME performance based on entrepreneur perceptions (Kiyabo & Isaga, 2020).Business performance determines the extent to which business target tasks are completed compared to the final results at the end of the business period.Business performance also depends on the success rate of the company in meeting the goals expected by the business owner (Zada et al., 2021)This means that every work result can assist owners or managers in making decisions related to the continuity and growth of their business.

Knowledge of SME performance
Adriana & Brahmayanti, (2010) conceptualize knowledge as a person's understanding of science and technology obtained through the learning process and experiences during his life.This emphasizes that an organization requires knowledge management to achieve financial success (Yan, 2018).Apart from that, standard performance will be achieved if it is accompanied by knowledge management and innovation (Khan et al., 2021).Empirically the research results explain the relationship between knowledge and SME performance (Hadiyati and Halim, 2023;Sedyastuti et al., 2021).Integration of knowledge management drives the SMEs in a more productive way is described by many researchers (Arafah et al., 2023;Fulgence et al., 2023;Giampaoli et al., 2024;Widodo, 2023) have been demonstrated.From the literature, the first hypothesis can be developed: H1: Knowledge has a positive effect on the performance of Small and Medium Enterprises in Padang City

Government policy on the performance of SMEs
The growth of SMES is highly dependent on strategies, policies, and government support so that SMES continues to grow and can overcome various crises and survive in the global business world competition.(Sulieman et al., 2016).Many encouraging research results regarding government policies on the performance of small and medium enterprises (Sedyastuti et al., 2021;Wibawa, 2018).Different researchers has described the positive effect of Government policies on the SMEs (Joo & Min, 2023;Mai et al., 2024).In order to evaluate the effect of knowledge on the performance of small and mediumsized enterprises (SMEs), an analysis is necessary.The hypothesis can be organized in the following manner: H2: Government policies have a positive effect on the performance of SMEs in Padang City

Government policies moderates the relationship between knowledge and performance of SMEs
The connection between understanding and awareness knowledge variables and SME performance has been extensively investigated by several previous scholars.Mardikaningsih et al., 2022 have shown that knowledge is related to the performance of SMEs, as was done by that knowledge competence must be owned by every SME.So that later they can have a strategy in marketing while findings from other researchers such as (Audretsch et al. 2023) also discovered the concept of the relationship.In addition, there are also assumptions from additional scientists, that the relationship between knowledge and SMES performance can have an indirect nature because there are difficulties in this relationship.Some other researchers assume that government policy must be able to moderate the relationship between knowledge and SME performance.So that it can form the following causal hypothesis: H3: Government policies moderates the relationship between knowledge and performance of SMEs in Padang City.

RESEARCH METHOD
The research design used is causal.The scale used is the Likert scale.The population in this study was all SMES in the city of Padang.Quantitative data is derived from primary data collected by sending questionnaires to 250 people owners or managers of SMEs.The number of questionnaires that were returned and filled out was 200.The sampling technique used was convenience sampling.Data collection methods are field studies and online questionnaires.Analysis technique with descriptive statistics which provides an overview of the weight gain of the question items, the interval value derived from the weighted average of respondents' replies to the question items.The data analysis technique uses the PLS-SEM program.According to Monecke and Leisch, (2012), SEM with PLS consists of three components of the structural model or inner model, measurement model or outer model, and weighting scheme (weight relation).
The performance variable of SMEs is the scope to which a person's understanding of science and technology is obtained through the learning process as well as experiences during his or her life.(Hunjra et al., 2021) This variable is also evaluated with 5 (five) statement items obtained from), namely 1) Business management knowledge, 2) Product knowledge or services, 3) Consumer knowledge, 4) Promotion, 5) Marketing strategy.
Government policy is also interpreted as A policy in developing industry, especially small and medium enterprises or industries by providing actual activities such as training frequency, capital access, and business partnerships.This variable is measured using 3 (three) statement items, the name is 1) Training frequency 2) Capital access 3) Business partnerships (Eniola, and Entebang, 2015).Knowledge in this context is defined as a process of learning and experience that a person goes through during his lifetime to understand science and technology.The indicators used to measure knowledge are adopted (Fulgence et al. , 2023).

Description of Respondents
The characteristics of the respondents discussed in this study are broken down by position, gender, last education, and length of service.Respondents by position showed that 53.87% of respondents were owners, while 46.13% were managers.From the status of ownership, it can be explained that there are more status as owners than managers.Respondent-based gender shows that 48.80% are male and 51.20% are female.It can be seen that the majority of owners or managers of SMEs are women.Respondents based on last education show that the highest number of respondents with a senior high school level of education is 47.47%, 36% bachelor's, and 16.5% master's.In conclusion, the majority of SME owners or managers have a Senior High School education level.It shows that formal high school education is still the most common level of education among SME owners or managers.Respondent-based years of service, the highest number of respondents who have worked under 5 years is 50.93% shows that many new SMEs have been established in recent years.while the lowest number is over 10 years of 11.2%,This may hinder the growth of SMEs due to lack of knowledge and skills in managing business.While respondents with a working period of 5-10 years are 37.8% of respondents.It can be seen that the highest number of respondents is the largest number of respondents with a length of work under 5 years.This demonstrates the high interest in entrepreneurship, the increasing number of people interested in starting up so that they need support for SMEs to grow and sustain.

Structural-Partial Least Square Equation Model (SEM-PLS)
The Structural-Partial Least Square Equation Model (SEM-PLS) is utilized in the data analysis method.Table 1 presents the analysis results, which include descriptions of variables, external models, and measurement models.

Measurement Models
The measurement model is reviewed to determine the construct step's validity and reliability.Three testing criteria need to be carried out, including convergent validity, internal consistency assessment, and discriminant validity (Sarstedt et al., 2014)

Validity convergence
Convergent validity means that a set of indicators represents one latent variable and the underlying latent variable.This representation can be demonstrated through unidimensionality which can be expressed using the average value of the extracted variance (Average Variance Extracted/AVE).The AVE value is at least 0.5.This value describes adequate convergent validity which means that one latent variable can explain more than half of the variance of its indicators in the average (Ghozali, I. & Latan, 2015).The majority, of the research used a loading factor limit of 0.70.An indicator can be declared to meet convergent validity and have a high level of validity when the outer loadings value is > 0.70, while the Average Variance Extracted (AVE) value is > 0.50 (Now, 2016).Based on the logic used for each indicator, an AVE value of 0.50 indicates that the construct can explain more than 50% of the variance between indicators.On the other hand, the AVE value < indicates that the mean item exhibits a higher error compared to the variance described by the construct (Hair, J. F., 2017) See Figure 1.Based on the results of the outer load value in the first run, it is known that several indicators have a value less than 0.7, so that indicator must be discarded (Figure 2).Then run it again to get the results as shown in Figure 3.

Model Fit Indicator -Goodness OF Fit
Many researchers are still debating the application of goodness-of-fit in PLS-SEM (Hair et al., 2017).TechniquePLS-SEM does not consist of globally accepted fit applications in theory testing and confirmation.On the other hand, several studies (Bentler and Huang, 2014) developed a suitability measure for the PLS-SEM framework.On the other hand, several others (Henseler et al., 2014) developed a novel Standardized Root Mean Square Residual (SRMR) to measure the squared difference that occurs between observed correlations and model-implied correlations to validate the model.Here, a value < 0.08 is considered suitable.PLS consistently presents a fair value that is used to assess model fit.The SRMR value is 0.117, which means >0.08, indicating that the model is not fit.See table 3.

Building Reliability: Composite Reliability (CR) and Cronbach's Alpha
CR values vary between 0 and 1, where a higher value indicates a higher degree of reliability.This value is interpreted like Cronbach's alpha.CR values ranging between 0.895 and 1 are very acceptable in exploratory research.However, a CR value < 0.6 indicates a lack of consistent reliability in the data (Hair, 2014).

Convergent Validity: Average Variance Extracted (AVE)
Convergent validity is defined as the degree to which the value is positively associated with the alternative item value.The researchers calculated the Average Variance Extracted (AVE) (Hair et al., 2017).
AVE is a popular method used to establish convergent validity for constructs.Ave is defined as the average of the squared loading values for the indicators associated with the construct (i.e., the sum of the squared loadings divided by the number of indicators).Therefore, AVE looks similar to construct similarity.Based on the logic used for each indicator, an AVE value of 0.50 indicates that the construct can explain more than 50% of the variance between indicators.On the other hand, the AVE value < indicates that the mean item exhibits a higher error compared to the variance described by the construct (Hair, JF, 2017).AVE is estimated as follows: AVE = K² / n K is the loading factor of any single item, and n is the number of items in the model.Table 5 presents convergent validity values based on AVE values.All AVE values appear to be >0.5, which leads researchers to conclude that the convergent validity of the model is met.Results obtained from running the first model results in knowing that there are 3indicators that are not valid product knowledge or services, business partnerships, and trends of return on assets.For this reason, these indicators are omitted in the model that is run for the second process.So the following results are obtained.6, while the loading factor is worth more than 0.6, so it can be concluded that all indicators are valid for measuring the construct.All latent variables (constructs) also have AVE values > 0.5 and Cronbach alpha values and composite reliability > 0.7, As a result, it would be reasonable to conclude that all indicators are reliable for determining each construct.

Discriminant Validity: cross-loading, Fornell-Larcker Criteria, and HTMT
In this study, the researchers also tested the discriminant validity (i.e., the degree to which items differed between certain constructs or concept measures) of the proposed model using three different criteria, namely, cross-loading, Fornell-Larcker and HeteroTrait-MonoTrait (HTMT) ratio.).Hair et al. (2017) stated that cross-loading is the first step used to assess the discriminant validity of all indicators.Table 7 shows that the cross-loading technique can fulfill all the requirements because the outer loading value of an indicator is higher than the cross-loading with other constructions (ie, the bold value).As shown in the table8, the square root value of AVE on the diagonal is represented as a bold value, higher than the correlation value between constructs (the corresponding row and column values).
This shows that all constructs are associated with their respective indicators compared to other constructs in the model (Hair et al. 2019;Fornell and Larcker, 1981)which indicates acceptable discriminant validity (Hair, JF, 2017).Furthermore, the correlation between all exogenous constructs is <0.85 (Awang, 2015)thus fulfilling construct validity.et al., 2015) and stated that this method cannot reveal the lack of discriminant validity that exists in standard research scenarios.They proposed an alternative method, namely, the HeteroTrait-MonoRrait (HTMT) correlation ratio, depending on the multitrait-multimethod matrix.Here, we assessed discriminant validity using HTMT.When the HTMT value >0.9 is compared to the HTMT0.9value of 0.9 or compared to the HTMT0.85value of 0.85 (Kline, 2023)discriminant validity is unacceptable (Gold et al., 2001).As shown in Table 9, all values are low compared to the recommended value of 0.85, indicating discriminant validity.

Direct Hypothesis Testing
Standard path coefficients indicate the strength of the relationship between exogenous and endogenous constructs.Therefore, knowledge shows a greater effect on performance than the influence of Government Policy on Performance.(See Table 10) Based on the results obtained, then: 1. Knowledge has a direct effect on performance 2. Government policies have a direct effect on performance

Coefficient of Determination: value of R²
The R² value describes the variance that exists in the dependent variable which can be explained based on the independent variable.Thus, a higher R² value increases the predictive ability of the structural model.All researchers must ensure that their R² value must be high so that the model achieves a minimum level of explanatory power (Chicco et al. 2021).(Cohen, 2013)states that the value of R² is said to be substantial if it is > 0.26 and has adequate power of more than 0.02.( Hair et al. 2019) noted that the value of R² is quite large if it is >0.65 and has an acceptable power >0.19.On the other hand,Signt et al. ( 2013) stated that the R² value must be >0.75 with acceptable power >0.25.Table 4. 18 describes the R² values for the structural model.As shown in Table 11, all R² values are very high, indicating that the model exhibits good explanatory power.Note that the variance described in the Structural Performance of the endogenous constructs is 0.631 (63.1%).

Effect Size F²
In this study, researchers also investigated the effect size (f²).The value of f² helps determine whether the exogenous latent construct shows a substantial, moderate, or weak effect on the endogenous latent construct (Gefen et al., 2011).Furthermore, Hair et al. (2017) stated that every change in the value of R² must be determined.Cohen, (2013) suggested guidelines for measuring f² values, which should be 0.35 (significant effect), 0.15 (moderate effect), and 0.02 (small effect).See table

Predictive Relevance (Blindfolding) Q2
This study uses a blindfold process to test the predictive relevance of the proposed model.Hair et al. (2017) state that the blindfolding process should only be used for endogenous constructs that have reflective measurements.When the Q² value > 0, it indicates that the proposed model exhibits predictive relevance for certain endogenous constructs (Hair, JF, 2017).Table 13 shows that all Q² values > 0, indicating that the model shows appropriate predictive relevance.Regarding the value of Q², Hair, JF, (2017)it is recommended to use values of 0.35 (significant effect), 0.15 (moderate effect), and 0.02 (small effect) as predictive relevance measures.The results showed that one of the endogenous constructs had moderate relevance.

Moderation Assessment
The moderating variable is the variable that 'moderates the effect of the predictor variable on the outcome variable (Awang, 2015)and according to (Hayes, 2013)Moderation plays an important role in many social science theories.This research focuses on the relationship between knowledge (predictors) and SEMs performance (outcomes) as a function of Government Policy (moderator).The results in Table 14 show that the use of Government Policy moderates (strengthens) the impact of knowledge on performance, so that H3 is rejected.Based on the findings of the first hypothesis test (H1), it can be concluded that the knowledge variable has a positive effect on the performance of SMEs in Padang City.This finding suggests that the greater the expertise, the better the performance of SMEs.Vice versa, It's important to note that the level of knowledge directly affects the performance of SMEs.The lower the knowledge, the lower the performance.So that the managers or owners of SMEs must pay attention to these factors to survive amid uncertainty.This is in line with the results of the study ( Hadiyati and Halim, 2023;Felício et al., 2012;Mayr et al., 2020;Sedyastuti et al., 2021) which proves that Knowledge has a positive effect on the Performance of SMEs.This explains that knowledge reflects Business management knowledge, Product knowledge or services, Consumer knowledge, and Promotion and Marketing strategies which have been assessed quite well by respondents as a guide for future decision-making.Furthermore, these findings are different from research shows that knowledge does not affect the performance of SMEs (Fulgence et al. , 2023).The implications of Knowledge and Performace in SMEs in different ways , for example, multi-tasking role and scarcity of resources, risk, data-driven decision-making processes, efficient communications; and strategic issues (Walsh et al 2023).In this case, the researcher concluded that the difference in results was because the respondent had not made maximum use of the knowledge he had to manage the business he was running.
From the results of theoretical studies of empirical studies, the findings of this investigation have proven that knowledge has a fairly good contribution to the performance of SMEs in the city of Padang.So the implementation of this knowledge must be applied immediately.Strategies that can be used to improve the performance of SMEs in the concept of knowledge is a focus on Business management knowledge, Product knowledge or services, Consumer knowledge, Promotion and Marketing strategy,

Government policy and performance of SMEs.
Tests carried out on the second hypothesis (2) conclude that the Government policy variable has a positive effect on the performance of SMEs in the city of Padang.Empirical facts show that the better the government policy, the expected performance of SMEs will increase.These findings reflect that government policies through training frequency, capital access, and business partnerships are expected to increase the competitiveness and performance of SMEs in the city of Padang.The results of this study are strengthened by (Eniola & Entebang, 2015;Hadiyati, 2015;Nkwabi & Mboya, 2019).These findings are different from those of Rahmi & Yuzaria, (2021) in the performance of a business is directly impacted by government policies but do not have a direct effect on the performance of SMEs, so government policies are still needed that are more focused on improving the performance of these SMEs.As well as (Y.Rahmat, 2022), that more and more government regulations are considered an obstacle to improving MSME performance.
Based on the theory and empirical studies above, if these findings are applied to SMEs in Padang City to improve the performance of SMEs, the steps that must be taken are to provide various kinds of training related to entrepreneurship, providing easy access to obtaining capital.as well as establishing partnerships with various parties to build business sustainability in the future.

Knowledge moderated by government policy on SMEs Performance.
According to the findings of the investigation, government policy as a moderation (strengthening) of the relationship between knowledge and performance is not supported.This means, even if the resulting government policies aim to optimize the performance of small and medium enterprises, whether they are getting better or not, this will not affect the performance of small and medium enterprises in Padang City, because government policies do not weaken the knowledge of the owners or managers of SMEs and do not strengthen performance SMEs in Padang City.

Managerial implication
Based on the results of this research, there are several managerial implications that can be applied by SME managers to improve their business performance by increasing the knowledge and skills of SME owners and managers through training and seminars on business management, product or service knowledge, consumer knowledge, and promotional strategies and marketing.Furthermore, providing access to sources of information and education and encouraging SME owners and managers to take part in formal training and education.From the government side, the policies that have been prepared support the development of SMEs, such as facilitating access to capital for SMEs, establishing cooperation between government, private and academic institutions to support the development of SMEs, measuring the effectiveness of policies in improving SME performance and making policy adjustments based on the results of evaluation and monitoring.The present research has emphasized and finds that influence of knowledge on the performance of small and medium enterprises and the government policies have a positive impact on SME performance.This findings proposes sustainably to the superior corporate business.

CONCLUSION
The research found that there is an effect an influence of knowledge on the performance of small and medium enterprises.Then, government policies have a positive effect on SME performance.And Government Policy as moderating (strengthening) the relationship between knowledge and performance is not accepted.Furthermore, conceptually, the study explains that the SME performance variables are varied enough that they have been adjusted to the criteria they have set.In addition, the research findings suggest that government policies as measured by indicators of training frequency, capital access, and business partnerships can be applied to moderate the relationship between knowledge and the performance of small and medium enterprises.
Although this research has several limitations, such as the research sample which only uses a sample of SMEs in Padang City, so the results of this research cannot necessarily be generalized to all SMEs in Indonesia.Apart from that, research data is only obtained from questionnaires, so there is the possibility of bias in respondents' answers.For variables, the research only examines several variables in influencing SME performance, namely knowledge and government policy as moderator variables, while there are still many other variables that have not been studied in this research.Even though there are still many limitations, this research is able to make an important contribution in understanding the performance of SMEs.Future research needs to be conducted to address the limitations of this study and improve our understanding of SME performance.

Figure 3
Figure 3 Outer Loading Values in the Second Run

Table 1 .
Concepts and Measurement of The Variables

Table 3 .
Fit Summary Table 4 presents all CR values and Cronbach's alpha > 0.7 which indicate that construct reliability is met, and the proposed model illustrates stability and consistency.

Table 8
Results values for discriminant validity by Fornell-Larcker

Table 9 .
Results of discriminant validity by HTMT

Table 10 :
Results of structural path analysis

Table 11 .
The coefficient of determination, R² value Exogenous build

Table 14 .
Results of the Moderating Effect Hypothesis