Science, Technology and Development  Volume 33 Issue 3, 2014

Research Article

Forecasting of Garlic Area and Production in Pakistan
Sobia Naheed
Social Sciences Research Institute, National Agricultural Research Centre, Islamabad, Pakistan.

Irum Raza
Social Sciences Research Institute, National Agricultural Research Centre, Islamabad, Pakistan.

M. Asif Masood
Social Sciences Research Institute, National Agricultural Research Centre, Islamabad, Pakistan.

Muhammad Zubair Anwar
Social Sciences Research Institute, National Agricultural Research Centre, Islamabad, Pakistan.

Nusrat Habib
Social Sciences Research Institute, National Agricultural Research Centre, Islamabad, Pakistan.

Present study is designed to forecast area and production of garlic in Pakistan, using past trends. Secondary data for the period of 1980-1981 to 2011-2012 were used to estimate future prospects of garlic area and its production in Pakistan, using (ARIMA) model. ARIMA (0, 1, 1) was appropriate form for estimation. The ARIMA model showed that forecast values of area and production would be 6.75 (000’hec) and 57.11 (000’t) in 2012 and 2013, respectively. If the present growth remains the same then the area and production of garlic crop would be 6.98 thousand hectare and 59.65 thousand tonnes, respectively in 2016 and 2017. The capacity of superior area and production lies in adequate accessibility of guidance to farming society, soil preservation and renovation and particularly the accommodating government strategies concerning garlic farming in the country. These forecasts will help the decision makers to make better policies regarding crop production, price and consumption.
    How to Cite:
Sobia Naheed, Irum Raza, M. Asif Masood, Muhammad Zubair Anwar and Nusrat Habib , 2014. Forecasting of Garlic Area and Production in Pakistan. Science, Technology and Development, 33: 123-126
DOI: 10.3923/std.2014.123.126


Garlic is a herbaceous plant belonging to the family Lilaceae with a botanical name Allium sativum which also consists of leeks, onion and shallots. It is perpetual with an underground bulb (head) with collection of pungent bulblets usually called cloves. On the basis of cultivation, garlic is the second most developed allium after onion. It is also renowned worldwide in medication for aliments of various physiological disorders and is also used as a precious spice in foods. It is grown everywhere in Pakistan and its production in 2011-12 was 1698.1 tonnes occupying a total area of 172.4 thousand hectare of Pakistan (GOP, 2012).

Pakistan is ranked among the top 10 garlic importing countries of the world from 2001-2010, with the exception of year 2001 and 2002, in which Pakistan is ranked as the 13th and 14th largest importer of garlic. During 2001-2010, Pakistan was ranked only twice among the top 20 garlic exporting countries of the world, in the year 2003 and 2008. In the year 2003 and 2008, Pakistan was the net importer, as well as, a net exporter of garlic (FAO, 2010).

Likewise, an increase in domestic production of garlic can be achieved by enhancing productivity of garlic crop. Productivity can be accelerated by introducing new technology or by improvement in efficiency or both. In Pakistan, the adoption rate of new technology is very slow, therefore, improvement in efficiency is an appropriate option to increase the agriculture productivity in short run (Javed et al., 2008). Measurement of the efficiency of agricultural production is an important issue in developing countries.

Keeping in perspective the vitality of garlic, it is important to estimate future area and production prospects. Normally ARIMA model has been utilised to inference area and production of major and minor crops and is, therefore, used in this study to estimate garlic area and production Forecasting techniques in agriculture consist of forecasting of production/yield, area of crops and forewarning of incidence of crop pests and diseases. Auspicious and dependable forecast give significant and convenient input for suitable, foresighted and notified arranging in farming which is full of lacks of determination. Forecast of crop production prior to harvest are needed for different arrangement choices identifying with space, dispersion, estimating, promoting, import-export and so on (Agrawal, 2010).

The aim and purpose of the present study is twofold, one is to check the past trends of garlic area and its production in Pakistan and second is to forecast area and production in the next five years using an Auto Regressive Integrated Moving Average (ARIMA) models. This study will provide help to farmers to identify factors that affect garlic growers’ technical efficiency and determining the opportunity for increasing output. The findings of this study will also be beneficial for policy-makers to form sound programmes related to expanding garlic production potential more effectively.


Secondary time series data for the period 1981-82 to 2010-11 were collected from various issues of Government Publications, such as, Agricultural Statistics of Pakistan and Pakistan Economic Survey. ARIMA model as proposed by Box and Jenkins (1970) was applied to forecast garlic area and production for the period 2012-13 to 2016-17. The model of ARIMA is denoted by (q, d, p), wherever ‘q’ is the sort of the moving average process, ‘d’ is the sort of the data stationary and "p" locates for the sort of the auto regressive process. The general form of the model (q, d, p) as explained by (Judge et al.,1988) is given below.


wherever, Δd indicates differencing of sort,

Δzt = zt-zt-1, Δ2zt = Δzt-Δt-1 and so forth, Zt-1 --- yt-p are past explanation, δ , γ1 – γ p are limitations to be guesstimated alike of the Auto Regressive procedure (AR) to regression coefficients of order "q" denoted by AR and is written as


wherever, et is estimation error,

Though MA model of organise p, MA tin be defined as:


The model was estimated using computer package "Minitab" stationarity of data were checked using normal probability plot and residual plots. Histogram plot was also drawn to check the shape of distribution. No outline observation was observed which confirmed a stationary series. Forecasts for the period 2012-13 up to 2016-17 was made using ARIMA (0, 1, 1) where 1 stands for difference (d), 1 for moving average (q) and 0 for autoregressive (p). Data were analysed in Minitab software version 15 (Minitab, 2007).

Results and Discussion

The area of garlic was 4.9 thousand hectares in 1980-81 that was minimum and production 5797.99 in 1981. The highest production of garlic in Pakistan was attained in 2011-12 (55.6 thousand tonnes) and minimum in 1980-81 year (36.9 thousand tonnes). After applying diagnostic checks a stationary series was obtained and an estimation of garlic area and production was made.

The model requirement implicated the plots of the auto correlation function (ACF), partial auto correlation function (PACF) and the plot of the distinction sequence. Auto correlation function designated the arrangement of the autoregressive apparatus ‘q’ of the model, though the partial correlation function granted a signal for the limitation p. The first step was to ensure the stationarity of the figures. The time series plan of area and production demonstrated a rising trend. Auto correlation function and partial autocorrelation function of both successions explained stationary as all of the lags were lying under the assurance confines. ARIMA (0, 1, 1) was chosen to be an appropriate model for estimation.

The ARIMA model was employed using four steps that are model specification, model estimation, diagnostic inspection and forecast.

Model estimation: ARIMA (0, 1, 1) models were expected using MINITAB computer package and evaluation of the models for the garlic area and production data are given in Tables 1 and 2.

Diagnostic inspection: For diagnostic inspection of the anticipated models, diverse diagnostic test were employed on behalf of whether these were suitably fitted or not. Goodness of fit of the models is given as under:

Residual analysis: One of the pointer of the suitably built-in model is that of sprinkled residuals in a rectangular form about the zero at parallel level. The time series plot of residuals of area and production data showed distributed trend, therefore, models were integrated accurately by residual analysis.

In favor of first familiarity test, plot of residuals for area and production figures of garlic proved to be fairly accurate in an instantly line screening familiarity, which is essential situation for regularity. The next normality test was to design the histogram of residuals. If the histogram proves regularity, the model is a fine fit. The histogram of residuals of garlic area and production chain confirmed regularity. Residuals alongside integral values of garlic area and production explained that there were no usual patterns originated representing kindness of fit.

Projection of garlic area and production till 2016-2017: Previous thirty years data of area and production of garlic were used for model for forecasting of area and production. ARIMA (0, 1, 1) were obtained on behalf of 5 year to the front and estimates for garlic area and production which are specified in Table 3 alongwith 95% assurance period values. For 2012-13 forecasts of garlic area was about 6.7523 and 59.6594 thousand hectares with lower and upper limits of 5.5187 and 7.9859 thousand hectares, respectively. A garlic area forecast for the year 2016-17 was 6.9898 thousand hectares with lower and upper limits of 3.5892 and 10.3903 thousand hectares, respectively. Forecasts of garlic production showed an increasing trend. For 2012-13, a forecast value of garlic production was about 57.1124 thousands tonnes with lower and upper limits of 46.0629 and 68.1620 thousand tonnes respectively. Garlic production forecast for the year 2016-17 came to be about 59.6594 thousand tonnes with lower and upper limits of 27.3986 and 91.9202 thousand tonnes, respectively. Similarly quite a few methods for classified particular cases of ARIMA models have been recommended by Box and Jenkins (1970). Makridakis et al., (1982) have conversed the techniques of categorised univariate models.

Table 1: Concluding estimates of area parameters.

Differencing: 1 regular difference

Number of observations: Original series 32, after differencing 31

Residuals: S S = 11.4834 (back forecasts excluded)

M S = 0.3960 D F = 29

In Table 1, the modified Box and Pierce (1970) statistic for garlic area, calculated above, for lag 12 is 4.9 which has the observed significance level 0.122. It indicates that it is non-significant at SL=0.05.

Table 2: Concluding estimate of production parameters.

Differencing: 1 regular difference

Number of observations: Original series 32, after differencing 31

Residuals: SS = 921.297 (back forecasts expelled)

M S = 31.769 D F = 29

The modified Box and Pierce (1970) statistics for garlic production calculated above, for lag 12 were 7.4 with observed significance level 0.040. It indicates that it is significant at SL = 0.050 in Table 2.

Table 4 shows the results of forecasts of garlic area during 2012-13 to 2016-17 in 000’t

Table 3: Forecasts for garlic area from period 2012-13 to 2016-17 (95% Limits) (000’hec)

Table 4: Forecasts for garlic production from period 2012-13 to 2016-17 (95% Limits)(000’t)

Conclusion and Recommendation
This study revealed that ARIMA model was the best for forecasting garlic area and production in Pakistan as area and production of garlic over the years is increasing. Proper use of latest technology and agricultural inputs can definitely enhance the garlic production. It will fulfill the basic food necessities of the country and its total produced area will enlarge in upcoming by renovation and maintenance actions. This projection will help the government to compose strategies with considered to qualified price formation, invention and utilisation and also to ascertain dealings among further countries of the world. We compared ARIMA model with different forecasting techniques, such as, linear trend, quadratic and exponential growth but the estimates found with ARIMA model were comparatively closer to the actual values. Therefore, we selected this model for justification.


Agrawal, R., 2010. Forecast of crop production. Indian Statistics Research Institute, Library Avenue, New Delhi.

Box, G.E.P. and D.A. Pierce, 1970. Distribution of residual autocorrelations in autoregressive-integrated moving average time series models. J. Am. Stat. Assoc., 65: 1509-1526

Box, G.E.P. and G.M. Jenkins, 1970. Time Series Analysis: Forecasting and Control. Holden-Day Publisher, Oakland, CA., USA., ISBN-13: 9780816211043, Pages: 575.

FAO., 2010. FAO Statistical Yearbook 2010. Food and Agriculture Organization, Rome, Italy.

GOP., 2012. Agriculture statistics of Pakistan 2011-2012. Ministry of Food and Agriculture, Government of Pakistan, Islamabad, Pakistan.

Javed, M.I., S.A. Adil, M.S. Javed and S. Hassan, 2008. Efficiency analysis of rice-wheat system in Punjab, Pakistan. Pak. J. Agric. Sci., 45: 95-100

Judge, G.G., R.C. Hill, W.E. Griffiths, H. Lutkepohl and T.C. Lee, 1988. Introduction to the Theory and Practice of Econometrics. 2nd Edn., John Wiley and Sons., USA., ISBN-13: 978-0471624141, Pages: 1064.

Makridakis, S., A. Andersen, R. Carbone, R. Fildes and M. Hibon et al., 1982. The accuracy of extrapolation (time series) methods: Results of a forecasting competition. J. Forecasting, 1: 111-153

Minitab, 2007. Minitab statistical software, version 15. Minitab Inc., State College, PA., USA., January 31, 2007.


Science, Technology and Development © 2021