PRODUCTION AND EXPORT OF ONION : TIME SERIES ANALYSIS

India is the second largest producer ofonions in the world after China. It enjoys 19% share ofthe global onion production. The annual average production is 12 lakh tones. The study has been undertaken to understand the effect of Area and Yield on Production of Onion and to study export trends of Onion. The production of onion has significantly increased over the past 30 years (1980-2012) but there has been lots of fluctuations specially from 1995 onwards indicating that production ofonion is not steady year overyear but sometimes increases and decreases. Such fluctuations may arise due to dependency of agricultural crops on rainfall and monsoons. Excess rainfall or less rainfall hampers the production of crops among otherfactors.


INTRODUCTION
The onion (Allium cepa) is used as a vegetable and is the most widely cultivated species of the genus Allium. A. cepa is exclusively known from cultivation and its wild original form is not known. Onion is produced and consumed not only in India but also throughout the world. Onion is classified as a vegetable and has special qualities which add taste and flavor to food. It is used extensively in Indian cuisine and culinary preparations both in cooked and raw form. Onion possesses very good nutritive and medicinal values. Onion is consumed by all classes of people-poor and rich and hence assumes a place of an essential item.
India is the second largest producer of onions in the world after China. It enjoys 19% share of the global onion production. The annual average production is 12 lakh tones. The major Onion producing states are Maharashtra, Kamataka, Madhya Pradesh, Gujarat, Bihar, Andhra Pradesh, Rajasthan, Haryana and Tamil Nadu. Maharashtra ranks first in Onion production with a share of 32.20% of the total Indian production. However, in terms of productivity, Gujarat ranks first with an area share of 22.9%. In Maharashtra Onion cultivation is primarily centered in Nashik, Pune, Ahmednagar, Satara, Sholapur and Dhulia. These regions are endowed with well drained, noncrusting soil required for onion cultivation. The production is taken in three seasons, i.e. Kharif (May-July to Oct-Dec.) Late-Kharif (Aug-Sep. to Jan-Mar) and Rabi (Oct-Nov. to April-June). Though onion is produced in three seasons, those produced in rabi season are only suitable for storage as the variety grown in this season has higher TSS, dry matter and more number of outer dried intact scales.

OBJECTIVES
The study has been undertaken with the following Objectives: • To understand the effect of Area and Yield on Production of Onion.
To study export trends of Onion

LITERATURE REVIEW
John .A (2011) in her study stated that the probable reasons for rise in Onion Prices was mainly due to failure in production of the Kharif Crop due to spread of fungal diseases like Purple Anthracnose and Purple Blotch among the Kharif onion saplings. Also erratic monsoons that caused water logging in the flat crop beds resulted in spread of these fungal infections among the saplings. The humid climate that prevailed from August worsened the situation. Though fungicides are generally effective against these fungal diseases, the heavy downpour made spraying ineffective. The end result was an unprecedented fall in the yield of Onion. Sharma. P et al (2011) in their study did a time series analysis of production of Onion and found that the trends in area and production of onions revealed that there is significant increase in onion production resulting in a rise in market arrivals.
However, due to unseasonal rains, production of onions declined by about 20% in three major growing states during 2009-10 and 2010-11. To some extent, this reduction in production was offset by marginally higher production in other states like Rajasthan. The magnitude of decline in production did not affect arrivals in the market very much. The astronomical increase in the prices of onions was a result of hoarding of the stocks in anticipation of a rise in the price and higher retailer mark-ups. Rise in prices was also partly due to reduction in MEP and consequent increase in exports duririg November 2010. Moreover, the crop situation was not predicted accurately and thus information on loss inproduction was not anticipated by market intelligence.
Tuteja. U (2013) in her study found that the wholesale price of onion in growing states was around Rs 45-50 per kg. However, retail prices in different places are Rs 70-80 per kg. The huge gap of Rs 25-30 per kg in wholesale and consumer prices implies a margin of more than 50%. After deducting cost of logistics, 10-15%, the difference in cost price and selling price is still high. A huge markup is taking place in the retail chain and traders are cornering huge profits. It seems that government agencies like NAFED are unable to efficiently monitor price rise regularly in the domestic market. Also, it did not take timely remedial action when there is a probability of a major shortfall in supply. So far, the government does not have any effective regulatory cell to monitor and foresee such abrupt increase in prices of essential foodstuffs with inelastic demand.

METHODOLOGY
The data has been collected from Secondary Sources which are various Government databases like NAFED, National Horticultural Board etc.. The data has been analyzed using Adarsh Journal of Management Research -Vol. : 7 Issue : 2 September 2014 econometric tools. The econometric tools used are Ordinary least Square Method as well as Various tests to check the stationarity of the data. Descriptive statistics have been used to analyze the Wholesale and Retail Prices of Onion.  Graph 1 indicates that production of onion has significantly increased over the past 30 years . It can also be seen that the graph is not a smooth line but has lots of fluctuations specially from 1995 onwards indicating that production of onion is not steady year over year but sometimes increases and decreases. Such fluctuations may arise due to dependency of agricultural crops on rainfall and monsoons. Excess rainfall or less rainfall hampers the production of crops among other factors. The Exponential or the fitted trend line (straight line) is drawn in the Graph and is found to be sloping upwards, and the exponent of'e' is our growth estimate which is found to be 5.5% (i.e; 0.055* 100). This process of calculating Growth rate is called the Chart Method Growth. Growth estimate helps us to know the yearly rate of Growth of a certain variable (production in this case). This rate can also help us forecast production for years to come. Area under Onion Cultivation: Like the production graph we see that area under onion cultivation has gone up over the past 30 years, but even this graph is not a smooth rising line but comprises of fluctuations. So the question arises as to why there are such fluctuations in cultivation area? The increase or decline in production of Onion maybe due to the fact that cultivation of a crop largely depends upon the price it fetches and when production increases the prices go down, thus leading to losses for the farmers. Under these circumstances farmers decide to produce less of the crop in the next year which again leads to a deficit of supply, leading to a rise in price. This induces farmers to produce more the next year. The growth estimate is found to be 4.3%. The above bar diagram shows that the yield / hectare of onions has increased over the past 30 years. The yield was highest during the year 2008 to 2009 (16260 kg/hectare) and lowest in the year 1997-98(9091 kg/hectare). Thus the difference between the highest and the lowest yield is 7169. From the value of the exponential we can see that growth estimate rate of the yield over the past thirty years is 1.1%. Also evident from the chart is that there has not been a substantial increase of yield over the past thirty years. Thus we can infer that there is still scope for improvement in the yield through use of better variety seeds, better technologies and proper fertilizers. Also, farmers need to be educated about correct proportion of fertilizers and pesticides to be used or else more or less of it will only ruin the crop leading to fall in production  The above table represents various tests done to check whether the Time series data of Onion is stationary or not. There are various methods that can be used to check the stationarity of a data. Now the question arises as to why thestationarity of timeseries is so important? Because if a time series is nonstationary , we can study its behavior only for the time period under consideration. Each set of time series data will therefore be for a particular episode. As a consequence, it is not possible to generalize it to other time period. Therefore, for the purpose of forecasting , such (nonstationary) time series maybe of little practical value (Damodar and Gujararti).

Results and discussion
One such simple test of stationarity is based on the so-called Autocorrelation function(ACF). The ACF at lag k, is denoted by pk and is defined as: pk = yk/ yO = covariance at lag k / variance If we plot pk against K the graph we obtain is knovra as the population Correlogram, but in practice we can only compute the Sample Auto correlation Function, pJc where pk = yk )'Q where yk = ZC^", -V and vo ...Therefore if we plotpA-against ic we get the sample correlogram.
In the given time Series data it was found that the Autocorrelation Coefficients were exponentially Falling (ACF starts at very high value and decline slowly) indicating that the series is non stationary but after first differencing the ACF at 11 lag hovers around zero (the solid vertical line represents the Zero Axis) indicating that the series has become stationary. Another way to check the stationarity of the data is to check the Q-statistic (developed by Box and Pearce) and the Augmented Dicky Fuller Statistic. In both these test for stationary we do hypothesis testing in case of Q stat our null hypothesis is set at HO : The series is stationary against the alternative hypothesis HI : The series is non stationary. But it should be noted that in case of Q statistic the null hypothesis is accepted if the p-value is set to be above 0.05 otherwise we reject it. So in the above table we see that the p-value at Q stat at level is 0.000 thus we reject our null hypothesis and accept our alternative hypothesis. But after first differencing the p-value of the Q-stat has increased and is above 0.05 for all the variables thus in this case we accept our null hypothesis which means that the series has become stationary after first differencing. Now in case of the Augmented Dickey Fuller Test we check the value of the absolute value of the calculated t statistic and our HO is set as the Series has a unit Root (in other words the series is non stationary ) against our HI : There is no Unit Root (or in other words the series is stationary). From the above table we see that the calculated t value at level (without differencing) has been found to be statistically insignificant as the p-value is greater than 0.05, but after first differencing the p-value has been found to be significant thus we can reject the null hypothesis and accept our alternative hypothesis.
Therefore from the various test for stationary we have found that a Macroeconomic data is non stationary in nature but becomes stationary after the first differencing. We have regressed Production with area and yield using Ordinary least square method and have obtained the above given results. From the given results we see that even though all the slope coefficients have been found to be statistically significant (p-value = 0.000) and also the set of data has found to be a very good fit with 99% variation in production in onion being explained by area and yield, but the Darwin -Watson(d-stat) value has been found to be very low at 0.849 thus indicating the presence of autocorrelation. Thus we can say that the situation exemplifies the problem of spurious, or nonsense regression which arises when the data is Non stationary Ay, = P, + p, AX, + P, +«, Now in the same given set of data we have again run a regression but in this case we have taken the first differences of all the variables to make it stationary. From the above set of calculations we see that after the 1 st differencing process when we run a regression using Ordinary Least square method the value of Darwin-watson (d-stat) has increased and is much more closer to two indicating presence of weak autocorrelation. The estimated slope coefficients of Area and Yield were found to be significant at 99% level of significance which means that as the total area under onion cultivation goes up by 1% on an average, production of onion would go up by 5%. Also if the yield per hectare increases by 1% on an average, production would increase by 13 %. Both of these variables have a positive relationship with the production of onion. The value of R2 of 0.95 means that about 95% of the variation in Production of Onion is explained by area and yield and the remaining 5 % is vmexplained. The high R2 value indicates that the sample regression line fits the data very well.