ANALYSIS OF THE EFFECT OF CONVERSION OF RICE FIELDS ON AGRICULTURAL PRODUCTIVITY IN EAST NUSA TENGGARA

The centrality of agriculture as one of the large sectors that sustain Indonesia's GDP does not make this sector free from various problems. The transfer of the function or conversion of agricultural land to nonfarm is a serious problem that occurs in Indonesia, especially in East Nusa Tenggara. The explosion of population growth and the shift of the economy to the manufacturing sector are the strongest reasons for the narrowing of agricultural land. The productivity of agricultural products is something that needs to be considered. This study uses descriptive analysis techniques and supported data from BPS. The results of this study showed no relationship between the phenomenon of transfer of agricultural land functions and agricultural productivity in East Nusa Tenggara.

production and maintain environmental quality (Mulyani & Agus, Potensi Lahan Mendukung Revitalisasi Pertanian, 2006). whether the production of rice which will become rice can affect national food security. To maintain food security, several efforts can be made, including controlling the conversion of agricultural land, creating new agricultural land, intensifying agricultural systems by applying technology that can increase agricultural production and maintain environmental quality (Mulyani & Agus, Potensi Lahan Mendukung Revitalisasi Pertanian, 2006). Whether the production of rice which will become rice can affect national food security. To maintain food security, several efforts can be made, including controlling the conversion of agricultural land, creating new agricultural land, intensifying agricultural systems by applying technology that can increase agricultural production and maintain environmental quality (Mulyani & Agus, Potensi Lahan Mendukung Revitalisasi Pertanian, 2006).
According to research conducted (Zuhdi, 2021) From report (Kementerian Pertanian, 2020), in 2017 rice and corn became the largest contributors to the agricultural sector in NTT with a total production of 1,900,651 tons. Changes in land use in East Nusa Tenggara occurred due to high population growth and resulted in an increase in settlements that were previously productive agricultural land (Angin & Sunimbar, 2021). The important role of the agricultural sector in Indonesia does not directly free the sector from various problems, one of which is the conversion of agricultural land to non-agricultural land. Land conversion that occurs in developing countries such as Indonesia is caused by structural and demographic transformations of the economy. The structural transformation of the economy occurs on the basis of more advanced agriculture, while the demographic transformation occurs on the basis of rapid population growth (Kustiawan, 1997) . According to agricultural land statistics in 2015 released by the Center for Agricultural Data and Information Systems, Secretariat General of the Ministry of Agriculture, this shows that the growth of rice fields in Indonesia is -0.17 percent, which means there is a decrease in the area of agricultural land. (Kementerian Pertanian, 2020). This is reinforced by research conducted by Mulyani and her colleagues in rice production centers in Indonesia which shows the conversion of paddy fields covering an area of 96,512 ha/year. The conversion of agricultural land covering an area of 100,000 hectares/year that occurs in Indonesia at this time can only be balanced with the opening of rice fields covering an area of 40,000 hectares every year. there is a conversion of paddy fields covering an area of 96,512 ha/year. The conversion of agricultural land covering an area of 100,000 hectares/year that occurs in Indonesia at this time can only be balanced with the opening of rice fields covering an area of 40,000 hectares every year. there is a conversion of paddy fields covering an area of 96,512 ha/year (Mulyani, Nursyamsi, & Syakir, 2020). The conversion of agricultural land covering an area of 100,000 hectares/year that occurs in Indonesia today can only be balanced with the opening of 40 rice fields.
Based on research conducted by (Harini, Ariani, Supriyati, & Satriagasa, 2019) Changes in agricultural productive land that are increasingly widespread have an impact on decreasing agricultural production, especially rice. (Putra & Ismail, 2017)His research shows that building houses, saving, buying houses and working capital are factors that cause the conversion of agricultural land under 0.5 hectares and is a serious threat to national food security. Study (Sasongko, Safari, & Sari, 2017)adding the strong influence of the private sector in the decision of farmers to release their land to be used as factories. In research conducted by (Gumilang & Kuspriyanto, 2017) factors that cause farmers to change functions are the level of education, price and land area.
The research was conducted in East Nusa Tenggara. The issue of border areas is very interesting, especially if it is associated with the conversion of agricultural land to the value of agricultural production, which will affect the economic development of the region. This research was conducted using quantitative methods based on published secondary data. Secondary data was obtained from the Central Statistics Agency of East Nusa Tenggara Province. The secondary data used in this study include: data on the area of agricultural land, the amount of corn production, the amount of soybean production and also data on the amount of rice production from 2012 to 2015. The data collection of this research was carried out by studying the literature. The analytical model used to determine changes in agricultural land and rice production is trend analysis. Linear regression analysis was conducted between agricultural land area and agricultural production to determine the effect of agricultural land area on agricultural production. Regression is done using the E-views program. In this case, the amount of agricultural production is the affected variable, while the area of agricultural land is the affected variable. Aim : 1. Analyzing the effect of paddy field area on agricultural production 2. Analyzing the effect of agricultural productivity on economic growth in NTT Benefit : 1. Provide knowledge and information about agricultural production and its effect on economic growth 2. As a material for consideration and additional literature studies for the government in making policies, especially in the field of agriculture

RESEARCH METHODOLOGY
A. Data lTypes land lSources 1. Data Type The data used in this study used secondary data. according to (Sugiyono, 2013)What is meant by secondary data sources are sources that do not directly provide data to data collectors, for example through other people or through previous documents that are taken from reports, research journals, magazines and other supporting library materials.

Data source
Sources of research data obtained indirectly through intermediary media, for example from the Central Bureau of Statistics, magazines, information or other publications. The data can be in the form of information from the Central Statistics Agency (East Nusa Tenggara), newspapers related to agricultural data and conversion of paddy fields. The data collected include: the area of rice fields, the amount of corn production, the amount of rice production, and the amount of soybean production.

B. Data analysis technique
This study uses statistical methods for estimation purposes. In the statistical method, the analytical tool that can be used is regression analysis using panel data. Regression analysis is a study of the dependence of a variable with other variables, with the aim of estimating / predicting population values based on certain values of known variables. (Gujarati, 2003)This research will use multiple linear regression equation with the model: = ( 1 1 , 2 2 , 3 3 ) + (1) = + 1 1 + 2 2 + 3 3 + (2) Where: = Area of Agricultural Land in East Nusa Tenggara 1 = Total Rice Production in East Nusa Tenggara 2 = Total Corn Production in East Nusa Tenggara 3 = Total Soybean Production in East Nusa Tenggara 1 , 2 , 3 = Coefficient of each Variable = Constant = Residue To lsupport lthis lmodel, lwe lalso lcarry lout lseveral ladditional ltypes lof ltesting, lnamely:

Test lchow
The lspecification ltest laims lto ldetermine lthe lpanel ldata lanalysis lmodel lto lbe lused. lThe lChow ltest lis lused lto lchoose lbetween lthe lfixed leffect lmodel lor lthe lcommon leffect lmodel lthat lshould lbe lused. lIf the results of this specification test show a Chi-square probability of more than 0.05, the model chosen is the common effect. On the other hand, if the Chi-square probability is less than 0.05, the model that should be used is the fixed effect. When the selected model is fixed effect, it is necessary to do another test, namely the Hausmann test to find out whether it is better to use the fixed effect model (FEM) or the random effect model (REM).

Hausman test
This test aims to determine which model should be used, namely the fixed effect model (FEM) or the random effect model (REM). In FEM each object has a different intercept, but the intercept of each object does not change over time. This is called time-invariant. Whereas in REM, the intercept (together) represents the average value of all intercepts (cross sections) and components represents the (random) deviation of the individual intercepts from the mean value (Gujarati, 2003).

Classic assumption test
The classic test that we will do consists of: a. Normality test The normality test aims to test whether in the regression model, the dependent variable and the independent variable are both normally distributed or not. Decision making with Jargue-Bera test or JB test, that is, if the probability is >5%, then the variables are normally distributed. If this assumption is violated, the statistical test becomes invalid for a small sample size. There are two ways to detect whether the residuals are normally distributed or not, namely by graphical and statistical analysis. The residual is normally distributed if it has a significance value > 0.05 and is not normally distributed if it has a significance value < 0.05. b. Heteroscedasticity test The heteroscedasticity test aims to test whether in the regression model there is an inequality of variance from the residual of one observation to another observation. Heteroscedasticity is a condition where the variance inequality of the residuals in the regression model. A good regression model requires the absence of heteroscedasticity problems. Heteroscedasticity causes the estimator or estimator to be inefficient and the value of the coefficient of determination will be very high. To detect the presence or absence of heteroscedasticity by ooking at the pattern of dots on the regression scatterplots. If the points spread with an unclear pattern above and below the number 0 on the Y axis, there is no heteroscedasticity problem (Priyatno, 2013). If the variance of the residual value between observations remains, then this condition is called homoscedastic. However, if they are different, it is called heteroscedasticity. A good regression model is a model that is homoscedastic. It is said that heteroscedasticity occurs if the probability value of OBS*R-square on the results of the heteroscedasticity test is ess than 5% or 0.05. c. Multicollinearity test The multicollinearity test aims to test whether the regression model has a correlation between the independent variables or not. A good regression model should not have a correlation between the independent variables. If the independent variables are correlated with each other, then the variables are not ortugal. Unortugal variables are independent variables whose correlation value between independent variables is equal to zero. According to (Gujarati, 2003), if the correlation coefficient between independent variables is more than 0.8, it can be concluded that the model has multicollinearity problems. On the other hand, if the correlation coefficient is ess than 0.8, the model is free from multicollinearity. Multicollinearity itself is a condition where between two or more independent variables in the regression model there is a perfect or near perfect inear relationship. (Priyatno, 2013).
To detect the presence or absence of multicollinearity by ooking at the Tolerance and VIF values. The smaller the Tolerance value and the greater the VIF, the closer to the multicollinearity problem. In most studies, it is stated that if the tolerance is more than 0.1 and the VIF is ess than 10, it can be concluded that in the regression model there is no multicollinearity problem. (Priyatno, 2013) d. Autocorrelation test Autocorrelation is a condition where there is a correlation of residuals for one observation with another observation which is arranged according to a time series. A good regression model requires no autocorrelation problems. The impact caused by the presence of autocorrelation is that the sample variance cannot describe the population (Priyatno, 2013).
Autocorrelation can be defined as the correlation between members of a series of observations ordered by time (for times series data) or space (cross-section data). The autocorrelation test aims to test whether in the inear regression model there is a correlation between the confounding error in the time or

T Uji test
The test of the value of the t statistic is a test of the significance of the individual parameters. The statistical value of t shows how far the influence of the independent variables individually on the dependent variable. The test of the t statistical value is also called a partial test in the form of a regression coefficient (Purwanto and Dyah, 2017). Ho : 1, 2, 3 = 0 Ha : 1, 2, 3 0 The t statistic is calculated from the following formula: t test : t = ️1 ️ (️1) Where : 1= parameter coefficient se (β1) = standard error coefficient parameter The t-test method is to compare the t-count with the table. If the t-count value > the t-table value, then Ho is rejected, which means that the independent variable (X) affects the dependent variable (Y), is the evel of significance and (nk) the degrees of freedom, namely the number of n observations minus the number of independent variables in the model (Ghozali, 2018). The test used to test the hypothesis of each independent variable owned. The nature of the researcher is to reject h0 and accept ha. What is seen is the value of t-statistics and probability values from the results of the regression test. It is said that the variable is significant if the prob t-statistic value of the regression test results is ess than (alpha)

F Uji test
This test is to see the significance of the model owned. What is seen is the value of F-Statistics and prob (F-Statistics) on the results of the regression test. The F test has 2 ways of reading: a. Goodness of fit (feasibility of the model): the model is said to be feasible if the prob value (F-Statistics) in the regression test results is ess than (alpha) 1%, 5%, 10% Simultaneous: if the results of goodness of fit are significant, then the independent variables jointly affect the dependent variable that we have (declaration).
The F statistical test basically shows whether all the independent variables included in the model have a joint or simultaneous effect on the dependent variable. H0 is the joint hypothesis that: 1,β2, 3.......βk is simultaneously zero. Decision making: Ho :β1,β2, 3 , .. = 0 , all coefficients are simultaneously equal to zero. Ha :β1, 2, 3, .. 0 , all coefficients are simultaneously not equal to zero.
Hypothesis testing is often called testing the overall significance of the regression ine which wants to test whether Y is inearly related to X1, X2, X3. The joint hypothesis can be tested using the analysis of variance (ANOVA) technique. ( 1− 2)/( − −1) If the calculated F value > F table then HO is rejected and H1 is accepted, meaning that there is an effect of the independent variables together on the dependent variable. On the other hand, if F count < F table, then HO is accepted and H1 is rejected, meaning that there is no effect of the independent variables together on the dependent variable. The statistical F test that measures the overall significance of the regression ine can also be used to test the significance of R², in other words the F-statistical test is the same as testing R² (Ghozali, 2018).

Coefficient of Determination (R²)
The coefficient of determination or goodness of fit is used to determine how far the model's ability to explain the variation of the dependent variable is. The value of R² is in the range 0-1. The higher the number, the better the model is made and vice versa. The coefficient of determination essentially measures how far the model's ability to explain variations in the dependent variable is. The value of the coefficient of determination is between zero and one. A small value means the ability of the independent variables in explaining the variation of the dependent variable is very imited. A value close to one means that the independent variables provide almost all the information needed to predict the variation of the dependent variable (Ghozali, 2018).

RESULTS AND DISCUSSION
East Nusa Tenggara has 22 districts/cities. The 22 regencies/cities owned are often grouped into their respective regions and regions to find out trends in and development and the amount of agricultural production in each region, which I present in the following table.

Panel Data Analysis
Based on the results of the model test, below we present the results of the regression using the fixed effect model. In the previous test, the model has not passed the classical assumption test, so the estimation results are not consistent and can be said to be biased: Table 3

. Panel Data Analysis
Source: data processed 2021 Based on the probability value then: -Variable X1 rice production has a value of 0.2921> 0.05, so it can be concluded that this variable is not significant to variable Y (rice field area). -Variable X2 corn production has a value of 0.9151> 0.05, so it can be concluded that this variable is not significant to the Y variable (rice field area). -Variable X3 soybean production has a value of 0.1067>0.05, it can be concluded that this variable is not significant to the Y variable (rice field area).

T Test (Partial)
The results of the partial test analysis show that each independent variable individually does not significantly affect the dependent variable. Here we describe each of the independent variables: -Rice Production The effect of rice production on the area of paddy fields showed insignificant results. This can be seen from the t-count value of -1.062461 and the probability of 0.2921 in the 5% significance evel.

-Corn Production
The effect of corn production on paddy field area showed insignificant results. This can be seen from the t-count value of 0.107039 and the probability of 0.9151 in the 5% significance evel.

-Soybean Production
The effect of soybean production on the area of rice fields showed insignificant results. This can be seen from the t-count value of 1.636788 and the probability of 0.1067 in the 5% significance evel.

F Test (Simultaneous)
Based on the results of the analysis using Eviews 10 software, the F-count value is 41.68806 and the F probability is 0.000000. In the 5% significance evel, the F test is significant. So it can be concluded that all independent variables together have a significant effect on the dependent variable. Variables of rice production, corn production, and soybean production together have a significant effect on changes in and area. And the model has also fulfilled the goodness of fit.

Coefficient of Determination
The influence of the independent variable on the dependent variable is 94.07% indicated by the goodness of fit value or the coefficient of determination of 0.940762 and 5.93% is explained by other variables outside the model.

CONCLUSION
If we write the final equation of the research model that I have done, it will become: LAND_AREAll= 5791.9042536 -0.00975876185367*RICE + 0.00112832894347*CORN + 1.35880817463*SOYBEAN + [CX=F] Information : Yll= Area of paddy field X1 = Rice production X2 = Corn production X3 = Soybean production /CX=F = Error Variables of rice production, corn production and soybean production which do not significantly affect the area of land I assume because productivity does not affect the conversion of agricultural and, therefore these variables cannot affect the area of rice fields.
I am aware that my research did not pass the classical assumption test, therefore our research has a model that is not BLUE (Best inear Unbiased Estimator). And even this research still has shortcomings in many aspects. I hope that the next researcher can improve the research that I do.