Tracking the Performance of Soyabean Production in Nigeria

Mevcut arastirma, Nijerya'daki soya fasulyesi uretiminin buyume performansini, FAO veri tabanindan elde edilen 1961'den 2017'ye uzanan zaman serisi verilerini kullanarak deneysel olarak incelemistir. Toplanan veriler, tanimlayici istatistikler, buyume modeli, kararsizlik indeksi, Hazell ayrisma modeli, Nerlovian modeli ve ARIMA modeli kullanilarak analiz edilmistir. Ampirik kanitlar, Nijerya'daki soya fasulyesinin uretim egilimindeki buyumenin, buyuk olcude uretim alani artisindan kaynaklandigini ve bunun ulkede surdurulebilir soya fasulyesi gida guvenligi icin elverisli olmadigini gostermistir. Ayrica, ulkede soya fasulyesi uretiminde en buyuk istikrarsizlik kaynaginin risk ve belirsizlik oldugu gorulmustur. Burada risk, incelenen alandaki soya fasulyesi ciftcilerinin alan tahsisi kararini olumsuz etkilemistir. Ongorulen uretim egilimi, ulkedeki soya fasulyesi uretiminin, uretimin ana itici gucu olarak bolgeden bir paradigma kaymasi gozlemleyecegini gostermektedir. Bu nedenle teknoloji, Nijerya'daki soya fasulyesi uretiminin ana itici gucu olacaktir. Surdurulebilirse egilim, ulkenin soya fasulyesi gida guvenligi icin bir donum noktasi olacak ve yuksek baski ve diger amaclar icin ekilebilir arazi icin rekabet eden talep nedeniyle araliksiz ciftci/coban catismalarini ve mulkiyet catismalarini azaltacaktir. Bu nedenle, soya fasulyesinin gelecekteki uretim egilimi teknolojikgelismeler tarafindan yonlendirileceginden, ekilebilir arazinin dortte birini diger tarimsal kullanimlara donusturecek politikalar memnuniyetle karsilanmalidir. Buna ek olarak, hukumet ciftcilerin ticaret suresini iyilestirmek icin ciftlik girdilerini subvanse etmede daha fazlasini yapmali ve boylece ureticilerin ekim alani tahsis kararini olumsuz yonde etkileyen riskleri azaltmalidir.


Introduction
In the last half-century, the global production of soyabeans has increased by a factor of eight to reach its present level of over 100 million metric tons per year. USA (45 percent), Brazil (20 percent) and China (12 percent) are the leading producers of this crop in the globe. Most of this remarkable growth was due, according to the FAO (no date), to the rapid rise in US production between 1950 and 1970, and the introduction of soyabean to Brazilian agriculture in the sixties.
Twenty-one African countries now grow soyabeans, with Nigeria having the highest 6-year (2000-05) average production of 486 000 tons over an area of 553 260 hectares, followed by South Africa with 205 270 tons out of 122 870 hectares, and then Uganda with 155 500 tons out of 139,500. Currently, Nigeria produces about 500 000 metric tons of soyabeans annually making it the black continent's largest producer of the commodity.
Approximately 11 million tons of soyabeans are consumed worldwide; annual consumption in Africa is about 618 000 tons, and another 4 800 tons of animal feed are used. Nigeria is sub-Saharan Africa's largest consumer of soyabeans, and then Uganda. Nigeria currently produces soyabean worth $85 million on the international market; while most soyabeans in the nation is consumed locally where it is used in soymilk production and specially formulated foods to support malnourished babies and children. The product's foreign market is rising, and sustainable.
Soyabeans development has received tremendous attention and direct government support in Nigeria in the last two decades (AMREC, 2007). The interest in promoting production and consumption of soyabean is related to the crop's tremendous economic and nutritional value. In addition, the crop has gained popularity in Nigeria since as far back as 1992; more than 200 000 hectares of land were devoted to cultivation. This was then the largest field of soya cultivation land in Africa as a whole (AMREC, 2007). Soya is extensively farmed, often by small-scale farmers, which accounts for its low yields. Despite this, there is a lot of hope to Nigeria's experiment in using soyabean as a food crop.
The current increase in soyabean development in Nigeria was due to several years of work from the mid-1960s to the 1980s, when scientists adopted a nationally organized soyabean research approach (Okoruwa, 2001). Soyabean cultivation in Nigeria has grown to a large extent over the years due to awareness of its economic advantages (Ojo, 2002). Due to increased knowledge among farmers about the versatility of soyabean and then the release of high-yielding varieties from research institutes working on soyabean growth in Nigeria, its production level has increased in some states.
As the largest producer of human and animal feed crops in West and Central Africa, Nigeria has a great potential to substitute certain imported vegetable oils for soy oil (RMRDC, 2004). Current domestic demand and home consumption have made the crop a versatile, multifunctional agricultural commodity that could be processed for human, livestock and industrial purposes in almost 365 ways. With the current ban on the importation of vegetable oils, some of the hitherto idle mills across the country are now looking inward, generating soyabean-based edible oils, preventing vegetable oil processing facilities from becoming inefficient and preventing inadequate supply of the oils. Therefore, it implies that the crop is an essential economic commodity that contributes significantly to the livelihoods of the farming population in the region. It is in the light of the above that this work aimed to analyze the sustainability efficiency of the crop with a view to ensuring soyabean food security in the country as it is one of the richest and cheapest protein sources.
The broad objective of the research is, therefore, to monitor the output of soyabean production in Nigeria, while the basic objectives were to analyze the soyabean production trend; to determine the growth pattern of soyabean production; to determine the status of instability in soyabean production; to determine the sources of instability in soyabean production; and to forecast soyabean production in Nigeria.

Materials and Methods
The study made use of time series data for the area, yield, production and prices which spanned from 1961 to 2017 (56 years), sourced from FAO database. The collected data were analyzed using descriptive statistics, growth model, instability index, Hazell decomposition model, Nerlovian's model and ARIMA model. For a detailed analysis of the growth trend and instability, the data were classified into periods based on the reform regimes that marked the economy of the country viz. pre-Structural Adjustment Period (pre-SAP) , SAP (1985-1999) and post-SAP (2000. In descending order, the objectives were achieved using descriptive statistics, growth model, instability index, Hazell's decomposition model, Nerlovian's model and ARIMA model 2.1 Empirical model

Growth rate
The compound annual growth rate calculated using the exponential model is given below: (3) Where, CAGR is compound growth rate; t is time period in year; is area/yield/production; is intercept; and, is the estimated parameter coefficient.

Instability index
Coefficient of variation (CV), Cuddy-Della Valle Index and Coppock's index were used to measure the variability in the production, area and yield of soyabean. Following Sandeep et al. (2016) andBoyalet al. (2015) the CV is shown below: The simple CV overestimates the degree of instability in long-term trends in time series results, while the Cuddy-Della Valle Index corrects the coefficient of variability by the index of instability as it de-trends the annual output and shows the exact direction of instability (Cuddy-Della Valle, 1978). Thus, it is a better measure to capture the instability of agricultural production and prices, and it is given below: CDII = CV*(1-R 2 ) 0.5 ………………………….. (5) Where CDII is the index of Cuddy-Della instability; where CV is the variance coefficient; and where R 2 is the multiple determination coefficient. In conformity with Dharke and Sharma (2009); and Debnath et al. (2015) the instability index was classified as low instability (≤15%) and high instability (>15%). Unlike CV, Coppock's instability index gives a close approximation of the average year-to-year percentage variation adjusted for trend (Ahmed and Joshi, 2013;Kumar et al., 2017;Umar et al., 2019) and the advantage is that it measures the instability in relation to the trend in production (Kumar et al., 2017). According to Kumar et al. (2017), a higher numerical value for the index represents agreater instability. Following Coppock (1962), the algebraic economic formula as used by Ahmed & Joshi (2013); Sandeep et al. (2016); Kumar et al. (2017); Umar et al. (2019) is given below: and, = ℎ ℎ

Sources of instability
Hazell's decomposition model Hazell's (1982) decomposition model was used to estimate the change in average output and change in the variance of output with respect to both regimes and the overall duration. Hazell decomposed the causes of change in the output average and modified the output variance into four (4) and ten (10) components as Umar et al. (2017 and quoted. Decomposition analysis of change in production assesses the quantum of increase or otherwise of production in year 'n' over the base year that results from change in the area, productivity or their interaction.
i. Changes in average production: Changes in the covariance between area and yield and changes in average area and mean yield are the causes. The model is shown below: ii. Change in variance decomposition: It is caused by ten factors and shown below is the model: ∆ ( ) ̅ 2 ( ) Interaction effect I (changes in mean area and mean yield) Interaction effect II (changes in mean area and yield variance) Interaction effect II (changes in mean yield and area variance) Interaction effect IV (changes in mean area and mean yield and changes in area-yield covariance)

Nerlovian model
The basic model that has come to be called as Nerlovian's price expectation model is as follows, following Sadiq et al. (2017).
= + * + …………………………………………………………….. (16) The hypothesis described in Equation (17) is the hypothesis of price expectation. The left-hand representation of this equation is the change of year-to-year demand forecasts. The expression on the right hand side is the error the farmers made in predicting the price during t-1. The price expectation coefficient (β) suggests that, during the current year, only a fraction of the last year's price prediction error is converted into revision in the expected price.
The magnitude of the adjustment to market and/or non-price adjustments is calculated by the "adjustment coefficient". The transition takes place in the previous year, in conjunction with the actual area planted. If the adjustment coefficient is one, farmers themselves will fully adjust the area under the crop during the current year and there will be 'no lags' in the adjustment. But if the adjustment coefficient is less than one, the adjustment will continue and cause lags, which will be distributed over time. The number of years required for 95 percent of the effect of the price to materialize is given below (Sadiq et al. 2017): In the present analysis, both the area under the crop's short-run (SRE) and long-run (LRE) elasticities with respect to price are calculated to analyze and compare the impact of price on the area's responsiveness in both the short-run and long-run. Below are the price elasticities: 2.1.6. ARIMA Box & Jenkins (1976) submitted that ARIMA (p, d, q), which is a combination of Autoregressive (AR) and Moving Average (MA) with an integration or differentiation order (d), denotes a non-seasonal ARIMA model. The p and q, respectively, are the autocorrelation order and the moving average (Gujarati et al., 2012). The order p Auto-regressive denoted as AR(p) is set out below: is the p-th autoregressive parameter and is the error term at time 't'. The general Moving Average of (MA) of order q or MA(q) can be written as follow: is the q-th moving average parameter and − is the error term at time 't-k'. ARIMA in general form is as follows: are values of past series with lag 1,………., p respectively. Modeling using ARMA methodology consists of four steps viz. model identification, model estimation, diagnostic checking and forecasting. Forecasting Accuracy Mean absolute prediction error (MAPE), relative mean square prediction error (RMSPE), relative mean absolute prediction error (RMAPE) (Paul, 2014), Theil's U statistics and R 2 were calculated using the following formulae for measuring the accuracy in fitted time series model: Where, 2 = coefficient of multiple determination, = Actual value; = Future value, and T = time period Figure 1 showed the trend patterns of production, area and yield across the sub-periods and the overall period. A perusal of the graph depicted an irregular trend for the production of soyabean over the entire period under consideration with area been the major cause of fluctuation as evident from the year 1969 to 2017; with a trend break in the year 2010. From the year 1961 to 1975, the production trend was flat due to mix of marginal changes in both area and yield; and thereafter, in the year 1976 to 1982, there was a slight incline in the production trend due to pronounced increase in the area cultivated under soyabean. During the latter period, the yield of soyabean was observed to be on the decline which may be attributed to policy inconsistencies owing to regime shifts; while infancy of the Nigerian economy owing to relegation of administration by the colonial master immediately after the country was granted independence which effected the managerial efficiency of the economy and the episodic factor (civil war) may be the reasons for the almost stagnant trend during the former period. It was observed that the production trend plummeted sharply in the year 1982 and maintained a flattened trend till the year 1984 which may be due to political crises viz. corruption by the political elites and weak international trade policies which affected the production trend as evident by the stagnant area and yield trends. The production trend suddenly inclined from the year 1985 till 1989, and then suddenly decline in the year 1990 and thereafter maintained a flattened trend till the year 1994. It was observed that the increase in the production trend from 1985 to 1989 is due to an increase in both area and yield with the increase in area been more pronounced than the incremental change in yield. Furthermore, the result showed sharp incremental changes in the production trend from the year 1995 and it persisted till the year 2008 with incremental changes in both area and yield been responsible. The incremental change in the trend of the area is more pronounced than the incremental contribution change recorded by yield. However, there was a deep decline in the production trend in the succeeding year (2009) due to a sharp fall in the area cultivated for soyabeaninspite of the high yield recorded in the same year. From the year 2009 to 2017, the production trend exhibited a zig-zag shape with more pronounced changes in the area than the yield been responsible for the upward and downward swings in soyabean production trend in the country. Summarily, from the year 1961 to 1994, the production trend of soyabean exhibited mild changes with the trend been flattened and thereafter, the changes sharply steep upward and downward. Incremental changes in area were more pronounced than the incremental changes in yield, thus the major source of change in the production trend of soyabean in Nigeria.

Trend pattern of soyabean production
A decomposition analysis across the three policy regime periods (Pre-SAP, SAP and Post-SAP) showed incremental change in yield to be the major source of increase in soyabean production during the pre-SAP despite that the production trend is flattened i.e. almost stagnant during the pre-SAP regime ( Figure 1b); incremental change in area is the major source of increase in the production trend during the SAP period( Figure 1c); while changes in both area and yield were the major source of incremental change in production trend during the post-SAP era ( Figure 1d). This general outlook did not come as a surprise as the pre-SAP era place more priority on the agricultural sector but the Nigerian's economy been nascent and poor in global trade integration affected the production of soyabean in Nigeria, thus the reason for the mild changes in the production trend of soyabean during this regime. For the SAP period, liberalization of the economy which led to the prioritization of black gold (crude oil) affected the agriculture sector: agricultural innovations and market promotion, thus the reason for area increase to sprout production increase rather than yield. For the post-SAP period, a decentralized and deregulated economy, competing demand and pressure on the existing land for purposes other than agriculture and the consequence of climate change due to exploitative human activities shrink the available arable land for agriculture production, thus the reason for the changes in soyabean production to be due to mixed change between area and yield during this regime. Figure 1. The trend patterns of production, area and yield across the sub-periods and the overall period.

Growth pattern of soybean production
A perusal of Table 3 showed the growth in the area, yield and production of soyabean in Nigeria to be magnificent over the three policy regimes considered. A significant increase in area was recorded during the SAP regime; a significant increase in yield was observed during the Post-SAP regime while a significant increase in production was recorded during the SAP and Post-SAP regimes. The annual average area under soyabean increased from 209 875 hectares during the pre-SAP era to 500 100 hectares during the SAP era, while the annual average yield sharply surged from 433 55kg during the SAP period to 919 62kg during the post-SAP regime. For the production, the annual average inclined steeply from 64208.33 tons during the pre-SAP regime to 221866.70 tons during the SAP regime, and thereafter geometrically increased to 538969.60 tons during the post-SAP regime. However, the growth in the area and yield were steady between the regimes that preceded post-SAP regime while that of the production was magnificent throughout the regimes.
The intra-year variation in area, yield and production of soyabean were observed to be considerably large as noticed from the variation in the annual compound growth rates across the three policy regimes and the overall period. The results showed the annual growth of production to have sharply increased from 0.1% during the pre-SAP era to 11.7% during the SAP era and then plummeted drastically to 1.9% during the post-SAP era. A similar trend was observed for the area as it steeply increased from 1.9% during the pre-SAP era to 5.2% during the SAP era and then sharply declined to 1.2% during the post-SAP regime. However, in the case of yield growth, unlike area and production growths it troughed (negative growth rate) during the pre-SAP regime (-1.8%) and thereafter revived steeply to 6.6% during the SAP regime, and then suddenly depressed to 0.6% during the SAP regime. Furthermore, the regime-wise analysis showed that the positive growth observed during the pre-SAP is solely due to increase in the area growth rate as the growth rate of yield was below the trough; while the positive production growth rates recorded during the succeeding regimes were due to both incremental growth rates in area and yield with yield and area growth rate increases been more pronounced during the SAP and post-SAP periods respectively. For the overall period, the instantaneous increase in the soyabean production growth rate (5.4%) is due to increase in both the growth rates of area (2.9%) and yield (2.6%) as their difference is marginal. Therefore, it can be inferred that instantaneous increase in the growth rate of area contributed more to the instantaneous increase in the growth rate of production across the regimes while the instantaneous growth rate of the duo (area and yield) interchangeably contributed to the incremental production growth rate observed for soyabean during the overall period. Therefore, the similar rise and fall in the growth rates of area and production as noticed across the policy regimes clearly indicate that the prevailing risk in the production of soyabean in Nigeria has been quite high.

Instability in soybean production
For SAP and the overall periods, the growth rates for the area, yield and production were quite impressive but the extent of variability in the area, yield and production were quite large as evident from the percentage of the coefficient of variation (CV) index. Though the growth rates for the area, yield and production were not that quite impressive during the pre-SAP and post-SAP regimes but the extent of the variability was comparatively very less in terms of the measurement parameters (Table 4).
The CV of the area, yield and production for the overall period (1961-2017) were 49.06%, 54.53% and 85.68% respectively, while the same for the SAP regime were 31.63%, 41.95% and 52.90% respectively. Therefore, it can be inferred that the higher compound growth rates in the area, yield and production during the aforementioned periods were accompanied by a greater degree of variability. Decomposition analysis across the sub-periods with regard to the CV showed that Nigeria was in the comfort zone for soyabean production during the pre-SAP and post-SAP as evident by the CV indexes of 14.73% and 17.40% respectively, while during the SAP and overall periods it was not in the comfort zone despite its impressive production performance in terms of annual growth as indicated by the CV values of 85.68% and 52.90% respectively. However, both area and yield interchangeably were the cause of high fluctuation during the SAP and the overall periods, but yield fluctuation was more pronounced in both periods.
The results of the Cuddy-Della Valle instability index (CDII) for the area, yield and production across the three regimes and the overall period showed instability to be low, thus indicating that the production of soyabean during the three regimes were in the comfort zone. In order words, the exact direction of instability in the production of soyabean using CDII showed a low instability during the pre-SAP period while the succeeding policy regimes exhibited high instability in the production of soyabean. A similar trend of high instability was observed in the production of soyabean for the overall period. While CV is the simplest measure of instability, it overestimates the level of instability in data from time series that is characterized by long-term trends. The Cuddy Della Valle Index de-trends the yearly data and shows the exact direction of uncertainty (Cuddy & Della Valle, 1978), thus, the appropriate measure to capture instability in agricultural production. Also, in examining the year to year percentage variation in the production of soybean, the Coppock's instability index (CII) showed high instability in the soyabean production across the regimes and the overall period.

Sources of change in soyabean production
Given that the trend pattern of soyabean production has been analyzed, thus, it's imperative to know the sources of the growth in the production of soyabean and the variance vis-a-vis year in year out. For the intra-regime-wise, yield effect was identified to the be the major source of production increase during the pre-SAP and SAP periods while area effect was found to be the major source of production increase during the post-SAP and the overall periods (Table 5). Furthermore, for the inter-regime-wise, the major source of change in the production increase between pre-SAP and post-SAP periods was area effect (55.92%); while yield effect (76.42%) was discovered to be the major source of production increase between the SAP and post-SAP regimes (Table  6). Therefore, it can be inferred that low level of technological development was responsible for the dominant effect of area for the former transition while high technological advancement viz. introduction of hybrids, improved varieties, technological breakthrough, mechanization and development of research institutes were responsible for the dominant effect for the latter transition. Also, for the overall period i.e across pre-SAP, SAP and post-SAP, the dominant source of change in soyabean production in the study area is yield effect. This implies that the country is aiming towards output expansion for food security viz. high productivity as available arable land is shrinking due to high competing pressure on land for many uses thereby causing loss of lives and properties as evident by the frequent reoccurring farmers-herders clashes and communal conflicts.

Sources of instability in soyabean production
The results of the decomposition analysis showed 'interaction between changes in mean area and yield variance effect' and 'interaction between changes in mean area and yield and change in area-yield covariance effect' to be the major sources of variation in the soyabean production between pre-SAP and SAP periods (Table 7). However, the effect of the former is marginally more pronounced than that of the latter. Between the SAP and post-SAP regimes, the dominant source of production variance is 'interaction between changes in mean area and yield and change in area-yield covariance' as it accounts for 70.21% variation in the total production variance of soyabean in the study area. However, for the overall period i.e. across the three regimes, the major source of soyabean production variance is 'change in residual' i.e. risk and uncertainty which accounts for 76.50% annual variation in soyabean production in Nigeria. This result is not surprising owing to the apprehension of capital loss by smallholder farmers who account for bulk production relies mostly on social capital due to lack of economic power and the heightened climate change effects.

Acreage response of soyabean farmers
A perusal of Table 8 showed the Nerlovian acreage response model to be the best fit as it satisfied the ordinary least squares (OLS) criteria. The diagnostic test results showed the residual to be devoid of autoregression/serial correlation as indicated by the Durbin-Watson (DW) statistic of 2.424 which is greater than the benchmark value of 1.5; and the autocorrelation Langrage multiplier (LM) test statistic value of 1.648 which is not different from zero at 10% degree of freedom. In furtherance of the above test, the variance of the residual did not exhibit a correlation (ei 2 ≠ej 2 ) i.e. has no arch effect as evident by the LM test statistic value of 2.27 which is not different from zero at 10% degree of freedom. This clearly indicates that the result is not a spurious or nonsense regression. In addition, the variance of the residual variable was found to be constant (homoscedasticity) as indicated by the Breusch-Pegan LM test statistic value of 7.02 which is not different from zero at 10% degree of freedom, thus the absence of heteroscedasticity. However, the residual variable did not exhibit a bell-shape (i.e. not normally distributed) as indicated by the Chi 2 test statistic value of 23.68 which is different from zero at 10% degree of freedom. Though, non-normality of the residual is not considered a serious problem as data in their natural form are not naturally distributed.
The test of multicollinearity indicates the absence of a collinear relationship between explanatory variables as evident by the explanatory variables variance inflation factors (VIF) which were below the VIF benchmark value of 10.0. The RESET test for specification test statistic value (3.71) not different from zero at 10% degree of freedom, implied that the structure of the model is appropriate i.e. rigid, thus suitable for prediction. It was also observed that there is no structural break in the equation i.e. the equation is stable as indicated by the CUSUM test statistic value of -1.007 which is not different from zero at 10% degree of freedom (Figure 2). In addition, it showed the population to be one and not subdivision(s). The coefficient of multiple determination is 0.8608, implying that 86.08% variation in the current acreage allocation is being influenced by the explanatory variables included in the model. In furtherance to the above, the R 2 value been normal indicates the absence of spurious correlation i.e. the estimated parameters does not drift with time trend. Thus, with these sufficient evidences it can be inferred that the model is the best fit and the estimated parameters are reliable for future prediction. The acreage allocation decision of soyabean farmers in Nigeria is influenced by lagged one year weather index, one year lagged yield, one year lagged cowpea and soyabean prices and one year lagged area cultivated under soyabean. Thus, the positive significant coefficients favorably influenced the area allocation decisions of the farmers while the negative significant coefficients negatively influenced the acreage allocation decisions of the soyabean farmers in the study area.
The negative significance of weather index implies that the effect of the one year lagged weather vagaries makes farmers decrease the current area cultivated under soyabean. Therefore, the marginal implication of an increase in weather vagaries will lead to a decrease in the current total area allocation to soyabean by 126393 hectares. In addition, the short-run (SR) and long-run (LR) elasticities implications of a unit change in the weather vagaries will lead to a decrease in the acreage allocation by 0.297% and 0.899% respectively. The positive significance of the one year lagged yield level indicates that high productivity owing to the adoption of improved varieties encouraged farmers to increase the current area allocation to soyabean in the study area. Thus, the marginal effect of a unit increase in the lagged year yield level by 1 kg will lead to an increase in the current area allocation to soyabean by 54.44 hectares. Also, the SR and LR elasticities implications of a unit increase in the lagged yield would lead to an increase in the current acreage allocation by 0.691% and 2.094% respectively.
The negative significance of the one year lagged soyabean price showed how fear of glut owing to cobweb effect decreased the current area allocation to soyabean. In addition, it implied high imperfection in the marketing of soyabean in the country. Therefore, the marginal effect of a unit increase in the producer price per ton will force farmers to decrease the current area allocation by 7.54 hectares. In addition, the SR and LR elasticities implications of a unit increase in the lagged producer price of soyabean would lead to a decline in the current acreage allocation to soyabean by 0.414% and 1.255% respectively. The positive significance of the one year lagged price of cowpea showed that soyabean farmers take advantage of the high price of the competing crop to increase their output supply in order to generate remunerative price i.e better farmers' term of trade. Thus, the marginal implication of a unit increase in the lagged producer price of the competing crop will lead to an increase in the current acreage allocated to soyabean by 3.22 hectares. In addition, the SR and LR elasticities implications of a unit increase in the lagged price of the competing crop will lead to a decrease in the current soybean area allocation by 0.201% and 0.608% respectively.
The positive significance of the one year lagged area cultivated under soyabean indicated partial adjustment or lesser rigidity in the current acreage allocated to soyabean in the study area. In addition, it implied that the producers of soyabean in the country not only considered the preceding year's price but also the past experience in the area allocation decision under soyabean. Thus, it can be inferred that the immediate lagged area allocation accounted for 67.04% of the current area allocation for soyabean production in the study area. In addition, the SR and LR elasticities implications of a unit increase in the lagged area allocation to soyabean will lead to an increase in the current soyabean acreage allocation by 0.656% and 1.989% respectively. The positive sign of the intercept coefficient though non-significant implies that acreage allocation to soyabean responded to technology i.e. technological improvements viz. improved seed varieties, agrochemicals and mechanization influenced farmers' decision on current area allocation to soyabean production. Thus, the marginal implication of a unit increase in the level of technology will encourage farmers to increase area allocation under soyabean production by 32075.10 hectares in the study area.

Short-run and long-run elasticities
A cursory review of the results showed that the farmers' response to a price change in the shortrun was inelastic (-0.414) while in the long-run they were elastic (-1.255) to price change as indicated by the coefficients of the former and latter which were less than unit and greater than unity respectively ( Table 9).
The short-run elasticity revealed soyabean farmers' acreage responsiveness to price changes during the preceding crop period. However, it should be noted that, as observed in many earlier studies, negative supply response is not an uncommon feature in the supply response. For example, in nearly six gram cultivating districts in Punjab, Sud andKahlon (1969) observed negative price coefficients; Cumming (1975) also observed a negative price coefficient in nearly half of India's 100 wheat-growing districts; Jhala (1979) also observed a negative price response in six of his fourteen groundnut-growing cases. This kind of conflicting estimate was reported in studies of Krishna (1963); Krishna & Rao (1967);and Bhowmickand Goswami (1998). This result is similar to the finding of Jain et al.(2005), who in his study on oilseeds output growth in Rajasthan State of India, observed a negative price coefficient for soyabean. Sadiq et al.(2017) in their study on supply response of cereal crops in Rajasthan State of India observed negative price coefficients for maize and Bajra. Of recent, Sadiq et al.(2019) in their study on the dynamics of root and tuber crops' acreage allocation and yield adjustment in Nigeria observed a negative price coefficient for yam.
If given the requisite time to adjust, the long-run elasticity represents the soyabean crop's acreage sensitivity to price changes. Because the soyabean crop showed very low long-run elasticity, the long-run effect of the price policy instrument on this crop would not have been significant. Because the soyabean crop showed very low long-run elasticity, the long-run effect of the price policy instrument on this crop would not have been significant. The number of years needed to materialize the price impact depends on the technical and institutional constraints for a specific crop that the farmers face. The higher the restrictions, the more time for adjustment is needed. It has been observed that the crop has taken substantial large time for adjustment to materialize its price. A similar result for soyabean crop was discovered by Jain et al.(2005) in their study on oilseed crops in Rajasthan State of India. The shorter the time for adjustment, the more efficient the market policy instruments are in bringing about desired improvement in a crop's supply (Sadiq et al., 2017). Also, it was observed that the farmers' responses to changes for the remaining parameters in the short-run and long-run were inelastic except for the lagged yield and area which were elastic.

Production forecast of soyabean
The future trend of soybean production in Nigeria was forecasted using ARIMA at different Autoregression (AR) and moving average (MA) levels (Table 10). All the variables viz. production, area and yield did not converged when he ARIMA was tested at level, but converged when tested at first difference, thus indicating that the variables were non-stationary at level but at first difference. Thus, the variables were stationary before the forecasts were made. A perusal of the Table showed ARIMA (1,1,1) and ARIMA (1,1,0) to be the best fits for forecasting the production and area; and, yield respectively, as they have the lowest Akaike information criteria (AIC) value among all the ARIMA levels estimated. Furthermore, all the best fit ARIMAs have their residuals devoid of autocorrelation and auto-covariance (arch effect) as evident by their respective autocorrelation LM and Arch effect LM tests respectively, which were not different from zero at 10% degree of freedom. However, the diagnostic test results showed that the residuals failed the test of normality as indicated by the Chi 2 statistic values of the respective ARIMAs which were different from zero at 10% degree of freedom. Just as earlier mentioned, non-normality of residual is not considered a serious problem as data in their normal form are not normally distributed. Therefore, the absence of white noise means that the soyabean production variables are predictable, thus good for policymaking, farmers' decision on supply allocation and consumption pattern. The forecasting ability of the variables were measured using the mean absolute prediction error (MAPE), root mean square error (RMSE), Theil's inequality coefficient (U) and the relative mean absolute prediction error (RMAPE) (Table 11b). Generally, the values of the RMAPE were very low i.e. less than 5% and also the values of U were less than 1.
The implication of the findings is, therefore, that the predictive error associated with the estimated equations in monitoring the actual data (ex-post prediction) was very small and negligible, and could therefore be used for ex-ante projection with strong forecast validity and accuracy. Shown in Table 11c and Figure ( [3][4][5] are the computed one-step-ahead out of the sample forecasts for production (ton), area (hectare) and yield (hg) of soyabean spanning from 2018-2027. A perusal of the graphs showed that there will be a gentle rise in the future production trend of soyabean; there will be a gentle decline in the future area trend of soybean, while the yield will be marked by a flat-slight trend rise during the forecasted periods. In case of any fluctuation, the forecasted figures will not go below the lower limit nor exceed the upper limit. Therefore, it can be inferred that the gentle rise in the future trend of soyabean production will be due to steady-flat rise in the yield. Thus, technology will be the major driver of soyabean production in Nigeria. The country is expected to witness a paradigm shift from area as the major driver of production to yield. The trend if sustained will be a major breakthrough for the country food security and containing the incessant farmers/herders clashes and tenurial conflicts owing to high pressure and competing demand for arable land for other purposes. Any policy which will convert arable land to other agricultural uses should be welcomed as the future production trend of soyabean will be driven by technological advancement. For sustainability of soyabean food security, viable policies that will sustain the quality of the soil viz. containing excessive exploitation of agricultural resources which add to global warming should be put in place and implemented holistically, otherwise, this forecasted trend will be jeopardized.

Conclusion and Recommendations
Based on the results, it can be inferred that the increased growth trend of soyabean production in Nigeria was majorly due to the increase in the area cultivated under soyabean. Furthermore, the rise and fall in the growth of area which is the major driver of production growth as noticed across the policy regimes clearly indicate that the prevailing risk in the production of soyabean in Nigeria has been quite high. In addition, evidence of risk and uncertainty accounts for the chunk (76.50%) of annual variation in soyabean production in the studied area as indicated by 'change in residual' parameter. Also, the farmers had a preference for risk with respect to the price of the competing crop but are risk-averse with respect to the price of the soyabean crop in taken decision on acreage allocation. Empirical evidence showed a good future production trend for soyabean in the country which will be driven by technological changes. Thus, viable policies that will sustain soil health status and the environment should be put in place. Besides, policies which will convert one-quarter of the arable land to other agricultural uses should be welcomed as the future production trend of soyabean will be driven by technological advancement. Owing to the production and marketing risks, government should do more in subsidizing farm inputs so that the farmers can have favorable farmers' terms of trade.