Crop Diversification in West Bengal:A District Level Analysis for the Period 1980-81 to 2011-12

This paper examines the dynamics of crop diversification in the pre and postliberalisation period addressing questions on the nature and scope of crop diversification in a small farm dominated economy like West Bengal and the factors influencing it. The analysis reflects the positive impact of operational area and cropping intensity on the degree of crop diversification. In contrast, factors like availability of irrigational facilities, degree of electrification, usage of fertilisers and price of output have a negative influence on the same in a particular region. Besides, a significant inter-district variation in the degree of crop diversification is noticed during the period under consideration. The study points out that a high level of diversification in an area does not essentially correspond to the district having a traditional resource base or a high endowment of modern inputs.


Introduction
India's adoption of New Economic Policy in the 1990's led to dramatic policy change from State intervention to that of market reform. IMF-World Bank directed market economic regime in the Indian economy in the early 1990's led to a drastic fall in food and fertiliser subsidy, promotion of private investments in agriculture, reduction in space for rural credit, priority sector lending and withdrawal of land reforms (Karmakar and Mukhopadhyay, 2007). Indian agricultural export sector became more open with import liberalisation on food, further deteriorating the situation. The proposals on development of Agricultural Export Zones (AEZs) implied a gradual drift towards market orientation and commercialisation of agriculture. The farmers lacked motivation to produce as they did not receive reasonable prices for their crop owing to high cost of production. Besides, government no longer had the authority to distribute the surplus foodgrains. Indian agriculture was, thus, going *Department of Economics, P.D.Women's College, Jalpaiguri, Email: smukherjee.m@gmail.com **Department of Economics and Politics, Visva-Bharati, Santiniketan, Email: soumyadip. chattopadhyay@gmail.com through a difficult phase with agriculture gradually becoming an unrewarding profession (Ibid).
Following such policy transformations, India has experienced crop diversification with fruits and vegetables in particular acquiring importance in the post-liberalisation period (Chand, 1996). In particular, demand for high value food commodities increased on account of sustained economic growth and rising tendency towards urbanisation which in turn facilitated the trend towards crop diversification (Rao, Birthal and Joshi, 2006). Change in the composition of consumption, as reflected through significant shift in the pattern of consumption away from foodgrains, induced producers to increase the production of high value crops (Gulati and Batila, 2001). World Bank (2005) observed that increased diversification in Indian agriculture resulted primarily from crop substitution rather than increased cropping intensity. Agricultural diversification in India was driven by factors like rising income, changing relative prices between cereals and high value agriculture, access to infrastructure and more open trade policy. Volatility in prices, together with yield and return risks, also acted as a major constraint in the way of changing the existing crop portfolio (World Bank, 2005). Dev and Rao (2005) categorically argued that the lack of an appropriate degree of price support to different classes of farmers can clearly be interpreted as a barrier to crop diversification.
On the supply side, high value agriculture is found to be more dominating in high rainfall areas with low levels of irrigation and mechanisation. Moreover, in these areas landholdings are relatively smaller, though labour endowment is higher, possibly because of cheap availability of family labour (Rao, Birthal and Joshi, 2006;Chakraborty and Kundu, 2009). Singh and Sidhu (2004) identified the availability of irrigation facilities at subsidised rate and the market support as the driving forces behind the shift towards crop diversification in Punjab. Institutionalised support also facilitates the opening up of new crop horizons. There is a need to support crop diversification drive based on the philosophy of marketisation and promotion of industries only under the conditions of improving technological and financial infrastructure (Chakraborty and Kundu, 2009). Joshi, Gulati, Birthal and Tewari (2004) observed a similar diversifying trend during 1990's in favour of high value commodities like fruits, vegetables, livestock and fisheries among the South Asian Countries with some intercountry variation, mainly on account of area augmentation. Much of the diversification in these countries came with negligible support from the government as the national policy goal was aimed towards achieving self-sufficiency in the production of foodgrains. The study further identified supply and demand side drivers of agricultural diversification. They are: technology adoption in terms of area under HYV foodgrains, fertiliser use, irrigated area and degree of mechanisation, rainfall, infrastructure in terms of market intensity and road length, relative profitability of high value commodities in comparison to other crops, proportion of smallholders,literacy rate in rural areas,urbanisation and per capita income. The study emphasised on the aspect of strengthening farm-firm linkages through appropriate institutional arrangements in the form of creation of cooperatives or contract farming system to ensure markets for such crops. This, in turn, would effectively link up the small farm holders with limited resources to the sources of seeds, fertilisers, credit, extension advice, mechanisation and guaranteed profitable markets for produce.
Among the Indian States, diversification in cropping pattern is a modern phenomenon in West Bengal which produces the largest quantity of rice in the country. The small farmers of the State are gradually diversifying in favour of high value crops like fruits, vegetables and flowers with little government support (Bhattacharya, 2007).
There are, of course,inter-district variations in the degree of diversification. High value crops being less costly and less water-intensive are easily affordable by the farmers of the State. One recent study on West Bengal found that the farmers in extremely backward regions, lacking irrigational facilities and having adverse natural conditions, resort to crop diversification as a means to protect their income and consumption requirements and in such situations high value crop production involves inefficiency, low productivity, low profit and negative returns (Mukherjee, 2015). However, in places wellendowed with irrigational and other infrastructural facilities, high value labourintensive crops generate higher income, relative efficiency and profitability by emphasising on intensive family labour. Against this background, this paper, using available secondary data for all the districts of West Bengal over the period 1980-81 to 2014-15, engages with the issues of crop diversification in West Bengal.

Methodology
The present research work was conducted by undertaking an analysis of crop diversification from1980-81 to 2014-15 across the 19 districts of West Bengal spread over diverse agro-climatic zones on the basis of secondary data collected from Statistical Abstracts and District Statistical Handbook, published by the Bureau of Applied Economics and Statistics of the Government of West Bengal (in short, "BUREAU") and Statistical Appendices to Economic Reviews of the State government. While some of our research questions and objectives could be addressed properly with an analysis of secondary data, a number of data gaps and information inadequacies confronted us. The objective of studying the dynamics of crop diversification over time could not be met with sufficiently available time series data over different periods. For example, due to non-availability of quantitative data on variables relating to infrastructure, marketing, storage, farm machinery, specific soil qualities, etc., our analysis of the factors influencing crop diversification behaviour had to rely, to a great extent, on qualitative information and wherever possible on quantitative data on this aspect.
In order to determine the extent of any variation in area allocation, production and productivity of crops since the early 80's in the State of West Bengal and to examine the dynamics of crop diversification in the pre and post-liberalisation period together with identification of the factors influencing crop diversification, we used the following statistical tools.

Growth Rate Estimation:
Having the time series of area, production and yield of different crops over a specified period, growth rates can be computed by several methods. The simplest method is to take the absolute or percentage difference in the levels of the first and final years divided by the number of intervening years. However, this method is not quite appropriate for measuring the growth rate of agricultural variables. In the face of wide year-to-year fluctuations, the computed growth rates from such an exercise would be heavily dependent on the choice of terminal years.
Other methods of computing the rate of growth consist of fitting linear or non-linear trends (such as exponential or second degree polynomial) which quite frequently give excellent fit to time series (which are not too long) data and work well for estimation of trends within the limits of the observed series. In our study, two alternative forms of regression models,viz., linear and exponential were attempted to find out which suits comparatively better to the available data on the basis of "goodness of fit" (in terms of the values of r 2 ). On the whole, the exponential form was found to be more appropriate. In particular, the following function is fitted to the variables of area, production and yield for computation of annual compound growth rates: Y= A (1+r) t = A.B t , Where, Y = the variable under study; t = time; A, B = parameters to be estimated. B = 1+r, r = annual compound growth rate. The form Y = A.B t .u is subjected to linear logarithmic reductions so that the regression model assumes the form: log y =log A + t. log B+ log u Where, u = a random error term. Having the estimate of regression co-efficient, this model allows for the computation of annual compound growth rate as: r = (b^-1) x 100, Where, b^ =Antilog of the estimated regression co-efficient.

Instability Analysis:
To study the nature of fluctuations in the variant values, we have calculated the Co-efficient of Variation (CV). This part of the present study is carried out entirely on the basis of secondary data collected from three major sources. These are Statistical Abstracts, and District Statistical Handbook, published by "BUREAU" and Statistical Appendices to Economic Reviews of the State government. The occasional data gaps have been bridged by using the figures for the corresponding years as made available in Economic Reviews of the Government.
Out of the numerous economic, social, demographic, institutional and other noneconomic variables influencing the farm efficiency and rational farming decisions, the most dominant ones are operational area of the crop, the yield rate and the level volume of production. These are also the most significant indicative variables on the nature and patterns of crop diversification as well as imply a causal Journal of Rural Development, Vol. 36, No. 4, October-December : 2017  xviii. Others = Sun hemp +Mesta +Cotton + Tobacco + Dry chillies +Ginger (Dry).
Out of the above crop scheme, the groups are as under: vi.
Other Cereals.
xx. All crops are not exposed to the regression analysis.
Nonetheless, we have considered all crops from (i) to (xix) for the general analysis. The collected data were subjected to a number of logical adjustments to conduct the statistical analysis. Firstly, the available time series spread over 35 years (1980-81 to 2014-15) was decomposed into the following sub-periods: A.
The sub-division (i) for general analysis gives us an opportunity to examine the earlier spurt in the process of agricultural growth revival in West Bengal which brought about a turnaround in the early 1980's. The sub-division (ii) helps us to capture the phase when the growth revival appeared to have gathered momentum. The next sub-division (iii) captures the impact of economic liberalisation during the early years, while sub-division (iv) incorporates the impact of introduction of WTO agreements.
The sub-division (v) is considered to examine whether there has been any saturation in the growth process following a good record during the earlier periods. Finally, the sub-division (vi) helps us to examine whether there has been any significant impact on the crop diversification dynamics in the State following the implementation of neo-liberal economic reforms even in a stronger manner than ever before. In fact, West Bengal has been one of the States which lagged behind the economic reforms process during the earlier periods and could somewhat pick up only of late. The last subdivision, thus, incorporates the effects of development of more recent origin including agribusiness, contract farming and rural market retail chains. The sub-divisions as under II provide us with an opportunity to study the problem of changes in cropping pattern over a period, capturing the agricultural growth resurgence of West Bengal (1980-81 to 1989-90) over another (1990-91 to 2013-14) that contains the impact of the neo-liberal economic reforms following GATT 2 negotiations and WTO 3 agreements.
The grand time series, 1980-81 to 2013-14, focuses on the long-run perspective of the problem. The second logical adjustment in the mode of presenting the data necessitated from the requirements of grouping the districts according to some standard norm, such as nature of actual variations in cropping patterns over time, such a problem must be avoided. In effect, the only possible way that remains is to consider an administrative sub-classification of districts. Specifically, we distributed 15 districts in our study according to the following classification (Table 1).
topographical, agro-climatic or geographical characteristics. Although, agriculture of the State can be represented in terms of well-defined agro-climatic zones, it involves the problem of overlapping of districts. Hence, segments of the same district may spill over among different agroclimatic zones leading to a double-counting problem. Since our purpose is to examine the Table 1 reveals that the cropping pattern in West Bengal is still dominated by foodgrains with 83.05 per cent area allocated under total foodgrains in 1980-81 and 66.2 per cent in 2014-15. Within this, the share of area under rice dominates (58.3 per cent in 2013-14). Since early 80's, following agrarian reforms and land redistribution programme, West Bengal experienced a revival of the small farm economy in rural areas. In fact, about 97 per cent of the operational holdings in the State are below 2 hectares in size. The high relative share of foodgrains is perhaps due to the impact of small farm size. However, in percentage terms,the area allocated under non-foodgrains has increased over time viz, area devoted to total oilseeds increased from 4.32 to 8.03 per cent while potato increased from 1.57 to 4.35 per cent during 1980-81 and 2013-14, respectively (Table   1). Fruits and vegetables started gaining significance in the post-liberalisation period with about 12.7 per cent area devoted to their cultivation in 2013-14.

Crop Diversification Trend in West Bengal: Growth and Instability Analysis
The declining tendencies of share of food crops over time clearly reflect signs of diversification. There is little evidence of change in area through substitution effect between rice and wheat. But within rice, substitution effect seems to be present through a shift from local to high yielding varieties. Since early 80's, HYV area increased through the boro component of rice and the penetration of irrigation through private shallow tube wells. Thus, there has been an area extension from 395.78 thousand hectares during the period 1980-81 to 1984-85 to 1279.18 thousand hectares during the period 2010-11 to 2014-15 under HYV variety of rice (

Table1: Administrative Sub-classification of Districts
also appears to be present in certain degrees. As observed in Table 2 As the crop pattern changes, the effect of the dynamics of crop-mix is expected to be reflected on agricultural production. Both production and yield rate of crops indicated a rise. As reported in Tables 3 and 4, it is observed that the production of rice increased from 6856.04 ('000 tonnes) during the period 1980-81 to 1984-85 to 14849.32 ('000 tonnes) during 2010-11 to 2014-15 while the yield rate recorded increase from 1323 kg/ hectare to 2762.4 kg/ hectare during the same period. The growth in production peaked up during 1980's mainly due to the yield rate effect. But, the yield growth could not be sustained during the 90's (Table 4). Similarly, for production of potato, the rise was from 2495.05 (in '000 tonnes) to 11544.6 (in '000 tonnes) and for yield rate it was from 19129.00 kg/hectare to 28806.6 kg/hectare during the same period.
We have also calculated the crop-wise growth rates of agricultural production in West Bengal (Table 5) by subdividing the entire period into three phases: Phase-I (1980Phase-I ( -81 to 1989 focusing on the phase of Agricultural Growth Resurgence of West Bengal, Phase-II (1990-91 to 2013-14) highlighting the impact of the Neo-liberal Economic Reforms following GATT Negotiations and WTO Agreements and Phase-III (1980-81 to 2013-14), emphasising on the problem from the long-run point of view. The results indicate that compared to the other two phases, the first phase reported remarkable growth rates of area, production and yield for most of the crops. For example, the total cereals grew at a rate of 6.41 per cent per annum during the first phase and the growth rates of area and yield were 1.16 and 5.2 per cent, respectively, implying that growth rate of yield was a significant contributor of the output growth.
The growth rate of production of total rice which includes all three varieties of rice grown in the State, namely, Aus or autumn rice, Aman or winter rice and Boro or summer rice declined from 6.91 per cent in the first phase to 2.57 per cent in the third phase (as reported in Table 5) owing to a fall in both growth rate of area as well as yield growth rate during the entire period taken into consideration. In fact, paddy crop reported the best performance regarding output growth both individually and in total in the first phase. The same picture also holds for the growth rate of output of cereal crops, which include all varieties of rice and wheat.
The output growth of production of rapeseed and mustard in the first phase could not be sustained primarily because of a reduction in area expansion to 2.8 per cent in the third phase together with a drop in yield growth to 1.39 per cent during the entire period. The growth rate of tea has however, improved marginally in the third phase to 2.09 per cent per annum mainly on account of increase in the growth rate of yield (1.39 per cent) over the first phase (0.46 per cent). Jute reported a decline in output growth on the whole as the marginal increase in area under cultivation (from -2.5 per cent in the first phase to 0.69 per cent in the third phase) has been outweighed by the declining growth rate of yield (3.99 per cent in the first phase to 1.86 per cent in the third phase as reported in Table 5).
Sugarcane experienced an enormous increase in production in the third phase (12.7 per cent) primarily caused by a substantial increase in growth rate of yield (from 2.33 per cent in the first phase to 12.2 per cent). The output growth of potato fell severely from 9.9 per cent in the first phase to 4.71 per cent in the third phase because of a fall in both growth rate of area (from 6.91 per cent in the first phase to 3.99 per cent in the third phase) and growth rate of yield (from 2.8 to 0.69 per cent, respectively, as reported in Table 5).
Fruits and vegetables occupied a significant position only in the post-liberalisation era indicating a trend towards diversification of cropping pattern with a growth rate of 3.04 per cent in the second and third phases (Table 5). It appears that the small and marginal farmers have resorted to crop diversification as a mechanism to avoid the risks associated with monocropping as well as to augment their income in order to accomplish their increased consumption requirements. On the whole, it is observed that the output growth of most of the crops declined significantly in the post-liberalisation period mainly as a result of decrease in both the area and the yield growth of most of the crops suggesting that the agricultural sector has been subjected to severe shocks in the postliberalisation period.

Nature of Instability:
We have complemented our analysis of growth performance with a study of the nature of stability with reference to area, production and yield of crops, including both aggregated variables and their decompositions at comparatively disaggregated levels under different phases (Tables 6 to 9).

Variability in Area, Production and Yield at the State Level:
Broadly, it is found that the area under total foodgrains was highly variable in Phase I as compared to that in Phase II (Table 6). Similarly, non-foodgrains recorded greater stability in the post-liberalisation period. The situation is quite similar in case of production and yield of different crops which reflects the relative stabilising effect of the post-liberalisation period possibly indicating saturation of opportunities in expansion of foodgrains due to the deflationary policy package followed in the 90's for the economy as a whole. However, the slow adjustment towards market-oriented high valued non-food crops is clearly indicated.

Variability by District Divisions:
The divisionwise variations in crop diversification point out to the fact that production conditions prevailing in such divisions are heterogeneous (Table 7). It appears from our study that the overall stabilising impact of Phase II in area under crops, e.g., as observed in case of Jalpaiguri division is an indication of physical and resource constraints towards crop experimentation in the new liberalised regime. This is particularly surprising since this division accounts for a considerable portion of crops like tea, jute, fruits and vegetables where the potential for diversification under new incentives appear to be high.
Interestingly, for Burdwan division (which accounts for a good part of traditionally advanced areas), it is seen that estimates of Coefficients of Variation (C.V.) have decreased for food crops from Phase I to Phase II. Presidency division appears to be at an intermediate level in this context. However, irrespective of the division, the variability in HYV summer rice has reported a decline in Phase II. With respect to yield of crops, it is seen that the Jalpaiguri division has become highly unstable in the period following economic liberalisation though an overall stabilising effect prevails for both Burdwan and Presidency divisions indicating a trend towards regional disparities subsequent to the New Economic Policy and reform measures.

Variability by Crop Clusters:
The estimates of C.V. indicate that the overall instability in area under crops has increased between Phase I (4.08) and Phase II (7.38) as reported in Table 8.
However, in case of foodgrains it increased only marginally in the second phase from 3.75 to 4.02. Similar is the case of non-foodgrains which reported a rise in variability from 9.13 to 10.69 possibly indicating an area adjustment and a trend towards experimenting with nonfoodgrains in the post-liberalisation era. However, non-foodgrains involve a very high degree of risk in yield fluctuations with variability increasing from 5.16 in Phase I to 11.61 in Phase II as reported in Table 8 which might act as a significant disincentive. This, in turn, points to the inefficiency of a systematic infrastructural intervention in agriculture in providing a positive stimulus to market-oriented diversification.

Inter-district Variation:
As revealed from Table  9, between 1980-81 and 1990-91, stability of area under crops point to a mixed pattern of rise and fall for the State as a whole. However, in the second phase, inter-district variations have gone up on the whole. In case of production, interdistrict heterogeneity has been on the rise in recent time periods for all crops except potato as compared to 1980-81. The variability of crops like rice, total foodgrains, total oilseeds and potato has decreased while the same has increased for crops like wheat, total pulses, other oilseeds, tea, jute and sugarcane. The yield of crops also reports similar facts. All these further corroborate the ineffectiveness of the neo-liberalised environment in agriculture towards reduction in inter-district gap, indirectly indicating the failure of the policy packages of New Economic Policy to provide positive incentives in the areas of comparative advantage of production.
In essence, our growth and instability analysis clearly points to the fact that growth rates of area, production and yield of most of the crops during the first phase (1980-81 to 1989-90) have been most noteworthy compared to Phase-II and III, viz., 1990-91 to 2013-14 and 1980-81 to 2013.The high yield growth rate of foodgrains during the 1980's led to the high growth rate of production. Crop pattern is still found to be dominated by foodgrains and within this the share of area under rice dominates. Growth rate of output of most of the crops was considerably high during 1980-81 to 1989-90 and in the subsequent period (1989-90 to 2013-14), the growth rates declined rapidly.
Instability in area under crops in case of foodgrains declined during Phase-II while that of non-foodgrains increased considerably indicating an area adjustment and a tendency towards crop experimentation in favour of non-foodgrains, though at a slow rate. During 1989-90 to 2013-14 (Table 9), the yield and the area growth rates of most of the crops had fallen leading to a fall in output growth rate. Both in cases of production and yield, there are clear indications of growing inter-district heterogeneity in recent time period. Yield fluctuations in case of non-foodgrains clearly indicate the inadequacy of a systematic infrastructural intervention in agriculture to motivate the process of market-oriented diversification.
We have constructed Simpson's Index 4 for crop diversification for 15 districts in West Bengal covering the period 1980-81 to 2013-14 as reported in Table 10. There is a significant interdistrict variation in the degree of crop diversification during 1980-81, 1990-91 and 2000-01 with the degree of variation recording a decline in 2013-14. This reconfirms the fact that agriculture in West Bengal is gradually on the way of being diversified across the space. Among the districts, Purulia is the least diversified district (with a crop diversification index of 0.24) in the State followed immediately by Birbhum (with a crop diversification index of 0.37 as reported in Table 10). Though the extent of diversification in cropping pattern increased in both the districts over time, yet it is quite negligible in comparison with other districts. The traditionally paddy growing district of Burdwan (with a crop diversification index of 0.39 as reported in Table 10), though considered to be one of the agriculturally advanced and enterprising district, is lagging well behind the districts of North Bengal in terms of diversification (with a crop diversification index of 0.53, 0.54, 0.7, 0.66 and 0.6, respectively, for the districts Dinajpur, Malda, Jalpaiguri, Darjeeling and Coochbehar). The same is also applicable for the district of Midnapur. An interesting feature is that diversification in Burdwan, though increased between 2000-01 and 2013-14, remains at a much lower level than the diversification achieved in Hooghly over time (with a crop diversification index of 0.65 in 2013-14 as reported in Table 10).
It is observed that the districts of North Bengal, namely Jalpaiguri, Darjeeling, Coochbehar, Dinajpur and to some extent Malda were diversified right from the initial years (with a crop diversification index of 0.5, 0.49, 0.43, 0.39 and 0.27, respectively, in the year 1980-81).
Among other districts, Hooghly was diversified from the beginning though the pace of diversification undoubtedly increased in the later half of the liberalisation era (crop diversification index increasing from 0.42 in 1980-81 to 0.65 in 2013-14). The same also applies to the districts of Nadia and Murshidabad (as reported in Table  10). In general, it is found that excluding the district of Hooghly all other districts belonging to South Bengal (namely Burdwan, Birbhum, Bankura, Midnapur, Purulia, 24 Parganas and Howrah) are lagging behind the districts of North Bengal in terms of diversification. However, there is no systematic way to conclude that a high level of diversification essentially corresponds to the districts having a traditional resource base or a high endowment of modern inputs. For example, diversification indices for the districts of Coochbehar, Darjeeling and Jalpaiguri compare fairly close with that of Hooghly, Murshidabad and Nadia (with a crop diversification index of 0.65, 0.63 and 0.68, respectively, for the districts of Hooghly, Murshidabad and Nadia as reported in Table 10). This indicates that crop diversification in resource constrained areas might be need-induced corresponding to a lowyielding basket of crops.

Factors Influencing Crop Diversification in West Bengal
Given the trend towards crop diversification, as evident from our analysis in the preceding section, it is imperative to examine the factors influencing such trend in West Bengal. This section employs tobit regression models to identify the determinants of crop diversification by different size classes of farmers on the basis of aggregated data of all the districts of West Bengal. In particular, Simpson Index of diversification has been regressed on returns. Hence, there is a positive relation between SI and average area operated under potato as greater the value of SI, higher is the degree of diversification and vice versa. As regards cropping intensity, it is also expected to influence crop diversification positively. Hence, the coefficient of cropping intensity has a positive sign. It appears that these variables have gained importance over time, particularly during the liberalisation period in influencing the degree of crop diversification (Joshi, Gulati, Birthal and Tewari, 2004). This is quite reasonable as both these variables incorporate the effects of the limiting most input of land being instrumental in crop experimentation.
However, the percentage of canal irrigated area contradicts our expectation as the estimated regression coefficient gives a negative sign in 1980. It remains with a negative sign except for the time reference 1990. In the context of West Bengal, canal irrigation potentials confront various constraints so that it is hardly favourable to crop experimentation and diversification. Normally, it is observed that the more is the percentage of area irrigated, the more willing is the farmer to undertake risk with new crops which ultimately has a positive influence on crop diversification. But, the percentage of area irrigated in the present model refers exclusively to canal irrigation which is mostly season and crop-specific. It is not fully under the control of the farmer and hence, is utilised mainly in the traditionally raised paddy crops during the kharif and summer seasons. The canal irrigation system runs on the basis of season, time-bound schedules and does not provide a uniform and even distribution of water throughout the year. Even within the prevailing time schedules of canal water, it is mainly biased in favour of kharif crops due to technological reasons. Hence, it seems that whatever canal irrigation is available that might be utilised by the farmers in favour of the traditional crops with a greater intensity in order to reap as much returns as possible out of them. Hence, canal water appears to be encouraging the farming of traditional crops which serve as a constraint to crop diversification and canal irrigation, which in general has a positive influence on specialisation in the main staple food crops of the area. As a result, the coefficient of percentage of canal irrigation is negative as a lower value of SI points towards specialisation rather than diversification. It is important to mention that the negative coefficient of percentage of canal irrigated area is found to be significant in the years 1980 and 2010. Canal irrigation provides cheap and assured availability of water. In general, traditional crops are more water-intensive. When there is lack of cheap availability of canal irrigation, the farmers are compelled to diversify to ensure a reasonable level of income. If there is proper irrigation facility, they do not find it necessary to diversify. This may be taken to imply that in West Bengal diversification is induced as a coping strategy to meet adverse conditions with regard to irrigation.
As regards fertilisers per unit of GCA, it is found to be negatively related with crop diversification index in the years 1980, 1990 and 2000, while it is positively related with crop diversification index in the year 2010. The negative sign of the estimated regression coefficient is found to be significant for the years 1990 and 2000. This apparently puzzling phenomenon may be explained in terms of the following possible factors: a.An increase in fertiliser use per unit of GCA might be mainly accounted for by increased area allocation under traditional crops which are more secure involving lesser risks in output compared to the crops which are newly introduced in the crop basket.
b.It might be that the increased dozes of FPGCA create adverse technological conditions in terms of soil quality and other factors which prevent the introduction of new crops. Thus, it is likely that the farmers find it profitable to concentrate on the fertiliser input more in the cultivation of stylised crops.
Our expectation that an increase in the number of electrified villages tends to produce a positive impact on crop diversification is also contradicted through the estimated regression coefficient which is found to be negative for all the time references used in the multiple regression framework. Moreover, these estimates are found to be highly significant for the years 1980, 2000 and 2010. This points out that the required power infrastructure is used by the farmers under constraints. For example, there might be problems of inadequate power supply to agricultural activities and use of electricity for unproductive consumption. Another thing is that the available power supply is utilised by the farmers primarily during the summer season (the time for raising a major component of the HYV paddy, the so-called 'Boro' variety) when canal water is not always available such that the remaining operative capacity to employ other mechanised instruments in non-paddy crops is reduced significantly.
Turning to the harvest price of potato as a determinant of crop diversification, quite contrary to our expectations, it is found to have a positive relation with SI except for the year 1990 where it remains negative. However, none of these estimates are found to be statistically significant. The positive relation between harvest price and SI might be explained through the inference that a rise in harvest price of potato can induce the farmers to diversify their crop baskets more in favour of cash crops other than potato to maximise their total returns. On the other hand, one possible reason behind the estimated inverse relation seems to be the fact that a higher harvest price promises a higher return on potato to the farmers which might induce them to specialise in potato as one of the dominating cash crops. Hence, a rise in harvest price might reduce the degree of crop diversification.

Summing Up
In the post-liberalisation period, the State of West Bengal, dominated mainly by the small and marginal farmers, has experienced a growing tendency towards diversification of cropping pattern. However, there is no systematic way to conclude that a high level of diversification essentially corresponds to the districts having a traditional resource base or a high endowment of modern inputs. Crop diversification in resource constrained areas might be need-induced corresponding to a low-yielding basket of crops. Diversification in cropping pattern is found to be influenced by factors like operational area, cropping intensity, availability of irrigation facilities, degree of electrification, usage of fertilisers, price of output, etc. Average area operated under potato and cropping intensity has a positive influence on crop diversification. Canal water appears to be encouraging the farming of traditional crops. Overdosing of the fertiliser input creates a negative impact on crop diversification while the power infrastructure used by the farmers under constraints primarily during the summer season for raising 'Boro' variety of paddy, too adversely affects crop diversification of a region. Similarly, harvest price of potato was found to have a negative impact on crop diversification in an area as higher harvest price induced specialisation rather than diversification.
It is true that successful implementation of the process of agricultural diversification in any region depends to a great extent on the technological, infrastructural and institutional developments as well as solution to problems related to land acquisition, prices, taxation, etc. In this context, government has recently introduced market-oriented dynamic policies to address the new emerging pattern of agriculture. But in order to ensure a higher growth path,macro policies should emphasise on upliftment of the State's agriculture by undertaking suitable programmes for bringing about changes in the cropping pattern. In an agriculturally advanced State of West Bengal, further growth prospects can be enhanced only through a regulatory rather than a promotional role of government.

1
For few selected crops, data on production and yield are expressed in units other than tonnes or kg/hectare.

2
GATT stands for General Agreement on Tariffs and Trade.
3 WTO stands for World Trade Organisation.