STRUCTURE OF PULSE PROCESSING IN INDIA Devesh Roy and Raj Chandra The pulse processing industry has the potential to play a special role in India because of the way pulses are consumed in the country and because of possible backward links that could uplift farmers living in marginal- ized environments. Pulse processing could also enhance the incomes of farm- ers and other participants in the value chain for crops that have not been among the best performers for a long time. In this chapter, we analyze the pulse processing sector’s growth and the relative roles of the organized and the unorganized sectors of the industry. We also identify the constraints facing pulse processing and suggest a way forward for the sector. Background A significant number of mills in India’s pulse processing sector remain part of the unorganized and small-scale manufacturing sector that uses traditional technology, while 75 percent of the pulses produced in India are processed in organized sector dal mills (Banerjee and Palke 2010).1 Although a large quan- tity of pulses are processed by medium-size industries, a significant amount is still processed in the rural sector without proper machinery. This not only affects the availability of dal in the rural sector due to loss during process- ing but also results in an inferior quality product that fetches lower prices. According to NAAS (2006), traditionally milled dal fetches 20 percent less in the market than the average quality dal and hence is generally sold in the rural market only. On the one hand, the government has promoted commer- cial food processing with several initiatives, deregulating and de-licensing it after 1991 reforms (except for alcoholic beverages). On the other hand, several 1 Dal is split pulses; it is the product that results after primary processing. The term unorganized sector in India refers to unincorporated private enterprises owned by individuals or households engaged in the sale or production of goods and services operated on a proprietary or partnership basis and with fewer than 10 total workers. Chapter 5 139 policy bottlenecks remain that constrain the sector. The pulse-milling sector, which earlier had been reserved for the small-scale sector, was “de-reserved” in the late 1990s. As such, no license or permission is now required for set- ting up a pulse mill, apart from permission from the Departments of Health, Industries, and Pollution Control Board. Regarding the impact and opportunities of global trade, policy has taken several turns over the past 25 years. The excise duty on food-processing items was removed in 1991, then reimposed in 1997, only to be removed again in 2001. In the food-processing industry, including pulses, the government gives automatic approval to foreign investment of up to 100 percent equity, except in a few cases; 100 percent export-oriented units (EOUs) are permit- ted to import raw material and capital goods free of duty. Moreover, in agro- based industries, EOUs are allowed to sell up to 50 percent of their products in the domestic tariff area (Dev and Rao 2004). The concept of food parks and agri-export zones (AEZ) has been initiated along with several incentive schemes (Dev and Rao 2004). Despite all of this, the value-addition in foods in India remains quite low today, measured at 7 percent in India as compared with 23 percent in China. Several policy-driven factors still inhibit India’s food-processing sector, government initiatives discussed above notwithstanding. According to the Department of Scientific and Industrial Research (DSIR), while the incidence of tariffs and indirect taxes has been reduced over the years, the tax structure for a range of processed foods (including pulses) is not uniform and not con- ducive to the processing sector. There is a different tax structure for branded and nonbranded food items, for example, and since branded items attract higher sales tax, they are costlier. Such imposed costs have adversely affected the processed food sector. Processed foods in India are costlier than fresh foods, unlike the situation in other countries. This stems from a series of taxes and duties applied in India. Most countries in the world do not levy taxes or duties on processed food products, seeking instead to promote value-addition in the food sector (India, DSIR 2007). Under the value-added tax (VAT), which is levied by state governments, most processed food products are taxed at varying rates of 1 percent, 4 percent, and 13 percent. Apart from VAT, other taxes (such as entry tax and octroi) are levied on food products. Also, the packing material attracts a high excise duty of 12 percent, further raising the costs of processed foods. Even the customs duty on packaging materials con- tinues to be high. Consequently, the net tax effect ranges from 21 percent to 23 percent on various food items (Dev and Rao 2004). 140 Chapter 5 Study Objectives and Data Sources Since the pulse processing industry in India has remained in the prelim- inary stage of development for some time, with a significant number of unorganized- sector mills, in this study, we first look at specific attributes of pulse processing in both the organized and unorganized sectors. We then dis- cuss the evidence for a changing relative share of the organized sector of pulse processing and examine growth rates over time. Next, we analyze some of the variables that may be undergoing a restructuring in the processing industry, such as scale of operation and capital-to-labor ratio. We conduct a regression analysis to identify the determinants of firms’ scale and productivity in the industry, which enables a glimpse into possible new growth areas. Toward the end of the chapter, we briefly look at the supply chain. The final section sum- marizes the findings and makes concluding observations along with some pol- icy implications. The analysis uses two secondary sources of data on the pulse processing sec- tor: the National Sample Survey (NSS) for data on manufacturing units in the unorganized sector, and the Annual Survey of Industries (ASI), collected by the Central Statistical Organization (CSO), for data on the organized (incor- porated) segment. The enterprises in the organized sector are those registered under the 1956 Company Act of India. One limitation is that the data from these two available sources are not disaggregated by type of pulse; instead, the data are combined for the pulse processing sector as a whole. The profile of pulse processing mills in the organized sector is analyzed using ASI unit-level data. For both NSS and ASI, the prescribed sample weights were used to arrive at the estimated population figures. For comparability across time, the mone- tary variables were deflated using the Wholesale Price Index (WPI) with base 2004– 2005. WPI better monitors price movements that reflect demand and supply in industry, manufacturing, and construction sectors and is used by the government in measuring inflation. Greater use of the consumer price index (with revisions) has become common only recently. Note that neither of these sources provides data for processing broken down by type of pulse; instead, the data cover aggregate pulse processing, which at a higher level is part of grain milling. Key Findings The evidence examined in this chapter shows that India has experienced an unambiguous scaling up in output per factory and in capacity— that is, in StruCture of pulSe proCeSSing in india 141 fixed capital per factory. This suggests that a production reconfiguration is going on involving new technologies and products at the factory level. Based on the data, overall the scale of individual pulse mills and the employment they generate remain quite small, not only in the unorganized but also in the organized sector. Moreover, the capital-to-labor ratios are lower in pulse pro- cessing than they are in most other food-processing sectors. For example, in 2006 the capital-to-labor ratio in sugar processing was more than 10 times greater than that in pulse processing (Bhavani, Gulati, and Roy 2006). Both as an engine of growth and for the upgrading of technology, it is the orga- nized sector that is likely to be the prime driver, with data showing that in rel- ative terms the organized sector has expanded in recent years. Measured by the number of mills, the unorganized sector remains comparatively large, but in terms of output there is a shift increasingly favoring the organized sector. Data also show that some states and regions are well ahead of others regarding the relative importance of the organized sector. The pattern seems to show that in poorer and more agrarian states, nearly all the processing is very small-scale and unorganized, while the organized and large-scale factories are clustered in urban areas. As a result, the most efficient pulse processing occurs in places far removed from most of the pulse-growing areas. The analysis assesses factory-level restructuring by examining changes in structural characteristics, such as scale of operation, technological change, and productivity.2 When we assess the factors that account for mills’ productiv- ity, we find strong evidence of state-specific factors playing a role. Moreover, there is mixed evidence for some agglomeration effects, such that where there is more processing activity (in terms of output but not in terms of the num- ber of mills) in a neighborhood, the average processing mill shows better per- formance. In terms of state-specific effects treating the leading state, Madhya Pradesh, as a benchmark, several states have significantly lower mill productiv- ity. Finally, we find that years of operation have a significant bearing on labor productivity, indicating that “learning by doing” could be playing a role. The Mechanics of Pulse Processing The processing of dal is the second-largest food-processing industry in India after rice milling and flour milling. The essential work of a dal mill follows 2 At a firm level, restructuring could be examined through mergers and acquisitions, but we do not have the data needed to analyze the issue that way, hence we focus on the industry- and fac- tory-level characteristics. 142 Chapter 5 three stages: dehulling the pulses, splitting them, and grinding into flour. According to the India Pulse Growers Association (IPGA), 75 percent of pulses produced in India are processed. Therefore, postharvest technology— whether it is traditional or advanced— plays an important role in the per capita availability of pulses. Pulse mills vary in size, from cottage industries to sophis- ticated factories with pneumatic conveyors. Nevertheless, most of the pulse processing industry is small scale, comprising thousands of dal mills distrib- uted throughout the country whose daily capacity is small, ranging from 0.5 ton per day to 10 tons per day. Measured in terms of both the number of mills and employment, a large part of the pulse processing is in the unorganized sec- tor; these mills use conventional technology with locally fabricated machin- ery. Geographically, pulse mills (of all sizes) are concentrated in the producing areas such as Indore (Madhya Pradesh), Jalgaon and Akola (Maharashtra), and in some big cities such as Chennai, Delhi, Hyderabad, Kolkata, and Mumbai. It is estimated that there are about 10,000 pulse mills in India, mostly in the private sector. On average they operate for 200 to 250 days per year. Primary versus Secondary Processing For pulses, primary processing consists of dehusking the grain, splitting it into dal, and grinding it into flour. Secondary processing refers to treatments that convert the dal and flour into acceptable, edible products (ICRISAT 1991). The processing follows these steps: The pulses are cleaned and foreign mat- ter, such as stones and mud, is removed. Next, the surfaces of the pulses are scratched, which improves absorption when they are next soaked in a mix- ture of water and vegetable oil (known as dampening). After they are dried (known as tempering), they are dehusked through grinding. Once the outer layer is removed, the pulses are split in half. These primary processes do not greatly influence the nutrient composition or acceptability of the pulses. Some additional practices like polishing with marble or leather polish, though not part of standard primary processing, can have health hazards or nutrition-de- pletion effects (see the discussion about polishing of pulses in this context in Chapter 7). Dals that are split— as in the case of pigeon pea, black matpe, green gram, and lentil— are more difficult to dehusk, so they require repeated operations by dehusking rollers. The soaking and drying mentioned above are repeated to loosen portions of husk that may be sticking even after repeated rolling. Linseed oil is often used to impart shine to the milled dal, which is appeal- ing to consumers. Some pulses—mostly chickpea, black matpe, and green StruCture of pulSe proCeSSing in india 143 gram—are milled to make flour (besan) through grinding. To give a better finish, some processors polish the dal. There is a belief that polishing leads to nutrient loss in pulses. Unpolished pulses are sold by one of the processors, Tata i-Shakti, as a differentiated health product because of this attribute. Secondary processing varies widely, depending on the consumers being tar- geted. One method involves soaking seeds in an alkaline solution to increase the water uptake of dal during cooking and decrease cooking time while increasing the dispersion of solids (Chavan et al. 1983). Unlike primary pro- cessing, elements of secondary processing do significantly influence the nutritional quality of pulse products. The processing time used, the tempera- ture, and the moisture level are three important factors in this regard. Moist heat methods are considered better than dry heat methods (Geervani and Theophilus 1980), and processing for longer than 10 minutes above 120°C is reported to cause considerable damage to proteins (Rama Rao 1974). This is quite important given that pulses are valued as a provider of protein. The secondary processing may involve a variety of dry or moist heating: roast- ing, boiling, steaming, or frying. Products using chickpea flour often require converting the flour into batter, which is then fried or else fermented and steamed. By controlling the proportion of water to flour in a batter, fried products of varied texture can be prepared (ICRISAT 1991). “Puffing” chick- peas, a secondary process, is a cottage industry in India. Processing Efficiency According to the IPGA, the output of the mills depends closely on the avail- ability of raw material, capital, and energy, as well as the capacity of each mill and the number of working days it operates. The different technologies involved in pulse processing have different levels of sophistication, depending on the properties of the grain and the efficiencies needed. Technology is likely the main differentiator between organized and unorganized pulse processing mills. To increase their use and consumption, in addition to their own pro- cessed variants, pulses are used as ingredients in other food items. Examples include breads, condiments, snacks, flours, gels, noodles, pasta, other baked goods, protein analogs, and snacks. The functionality of pulses as ingredients varies by the type of pulse milled and the type of milling procedure employed. The functional properties relevant to their use as ingredients include such attributes as water absorption capacity, oil absorption capacity, taste, texture, cooking time, and color, among others. 144 Chapter 5 The efficiency of pulse processing varies greatly according to the meth- ods used. According to Parpia (1973), domestic small-scale milling processes give yields on the order of 75 percent from chickpea and 68 percent from pigeon pea, whereas improved milling technologies give yields of more than 80 percent, with a theoretical maximum of 89 percent. According to Banerjee and Palke (2010), to minimize losses in processing the dal, industry should maximize the use of improved dal mills, which are highly versatile, technol- ogy savvy, and more energy efficient than the traditional mills. These new and improved dal mills have a dehusking efficiency of approximately 95 percent, while their split-pulse yields run between 80 percent and 85 percent, largely depending on the variety of the pulse and the conditioning of the pulse grain. Many agricultural universities and institutions recognized by the Indian Council of Agricultural Research (ICAR) have played a large role in devel- oping such improved dal mills.3 Today, new options have become available in food processing, including technologies for processing whole pulses, tech- niques for fractionating pulses into ingredients that preserve their functional and nutritional properties, and other potential applications to incorporate pulses into new food products. An Economic Analysis of the Organized and Unorganized Sectors of the Pulse Processing Sector As mentioned earlier, a salient feature of the Indian food industry in general, and its pulse processing sector in particular, has been the preponderance of the unorganized sector, which consists of numerous small units. At the same time, the organized sector of the total food-processing industry has been steadily growing for more than two decades, increasing from 64 percent of the total output in 1984– 1985 to 81 percent in 2000– 2001, and it is expected that it has grown further since (Bhavani, Gulati, and Roy 2006). An important cor- relate of the increasing formalization of the food-processing sector is a greater incidence of contractual relationships between farmers and processors (Gulati, Joshi, and Landes 2008). The policy reforms made in India in the early 1990s, such as de-licensing and dereservation for small-scale firms, and those made in 3 Examples of these institutes are Panjabrao Deshmukh Krishi Vidyapeeth (PDKV) in Akola, Central Food Technology Research Institute (CFTRI) in Mysore, and Central Institute of Agricultural Engineering (CIAE) in Bhopal. StruCture of pulSe proCeSSing in india 145 the 2000s, such as the launch of mini food parks, in combination with con- sumers’ changing incomes and tastes, were expected to encourage the indus- try’s organized sector, including the pulse processing segment.4 However, although organized pulse processing dominates in output and has increased its fixed assets over time, comparative data show that the unorganized sector remains dominant in generating employment, although the converse is true for its output and share in sales. In fact, the share in output of the unorganized sector has declined from an already low 26 percent in 2001 to just 23 percent in 2010 (Table 5.1). Employment Table 5.1 presents the comparative picture for the organized and unorga- nized pulse processing sectors in India. Measured by the number of mills, nearly 85 percent of pulse processing is in the unorganized sector. Measured by employment, the unorganized sector also has a large total share of employ- ment, at 70 percent (although this last number might be a bit overestimated, because employment in the unorganized sector includes part-time workers). The large share in total employment is partly due to the technology used, but it could also be policy-induced (for example, relating to labor market regula- tions). It is also a function of the number of unorganized mills. Hence, even though a typical mill in the unorganized sector employs fewer people than one in the organized sector because the number of unorganized mills is compar- atively large, the total employment in that segment is higher. In addition, the traditional manual technologies that most of the unorganized units operate with may require them to employ more workers. In contrast, organized mills 4 Under compulsory licensing under the Industries (Development and Regulation) Act of 1951, industries were put under compulsory licensing on account of such factors as environmental, safety, and strategic considerations. Furthermore, the industrial policy of small-scale reserva- tion forms a significant aspect of India’s industrial policy. Dereservation of such items is under- taken by the government at periodic intervals. All undertakings other than the small-scale industrial undertakings engaged in the manufacture of items reserved for manufacture in the small-scale sector are required to obtain an industrial license and undertake an export obliga- tion of 50 percent of the annual production. The exceptions to licensing are for undertakings operating under 100 percent Export Oriented Undertakings Scheme, the Export Processing Zone (EPZ), or the Special Economic Zone Schemes (SEZS). The government of India, in order to promote food processing, has chosen two schemes: namely, mega food parks and mini food parks. The former is based in larger areas between 50 to 100 acres, depending on the business plan, and have central processing centers combined with primary processing centers and collec- tion centers. Mini food parks are with much smaller areas allocated (30 acres) and need not have primary processing centers or collection centers. The extent of subsidy (in the form of financial assistance) is also lower for the mini parks. 146 Chapter 5 TAbLE 5.1 Characteristics of the unorganized and organized pulse processing sectors over time, 2001–2010 Characteristics 2001 2003 2005 2010 Unorganized Organized Unorganized Organized Unorganized Organized number of mills 7,897 900 8,496 1,330 8,034 1,517 total number of workers 23,830 10,773 27,812 13,688 29,058 12,642 number of men 17,309 7,321 24,427 9,186 24,785 10,027 number of women 6,521 3,452 3,385 4,502 4,273 2,615 percentage of men in total workforce 74 68 88 67 85 79 percentage of women in total workforce 26 32 12 33 16 21 average number of worker per mill 3.01 11.97 3.42 10.29 3.61 8.33 fixed capital per mill (rs millions) 0.05 1.44 0.09 0.88 0.33 3.37 gross value-added per worker (rs millions) 0.37 0.17 0.37 0.15 0.09 0.98 gross value added per mill (rs millions) 1.12 2.01 1.22 1.58 0.31 8.13 Capital-to-labor ratio 0.018 0.12 0.029 0.09 0.09 0.40 total output (rs millions) 13,760 38,731 43,569 77,010 81,480 266,152 Share in total output (%) 26.21a 73.78 36.13 63.86 23.41 76.48 Share in gross value-added per mill (%) 35.78 64.21 43.57 56.42 3.67 96.32 Share in gross value-added per worker (%) 68.51 31.48 71.15 28.84 8.41 91.58 Source: authors’ calculations based on data from annual Survey of industries and national Sample Survey organization (nSSo). Note: a the percentage share of the organized sector and the unorganized sector has been calculated by combining the values for 2001 and 2003 to match the data for the organized sector and the unorganized sector to the nearest time period available. StruCture of pulSe proCeSSing in india 147 operate with technologies that are more capital- and skill-intensive and there- fore less labor-intensive.5 This is reflected in much lower rates of capital per worker as well as capital employed per mill in the unorganized sector. Overall, pulse processing does not seem to be a big employment generator, with employment values of just 3.6 workers per mill and 8.33 workers per mill in the unorganized and orga- nized sectors, respectively. In terms of gender composition of employment, the organized sector actually has a much greater share of women in the workforce. Moreover, the scale is much higher in the organized mills as captured in terms of capital per mill and output per mill. Capital Table 5.2 presents the dynamics in the organized pulse processing sector over a recent nine-year period. Over that period there was a 70 percent rise (that is, a 6.1 percent compound annual growth rate) in the number of mills in the organized sector, and this was accompanied by substantial capital deepening as reflected in the growth in fixed capital. The growth in fixed capital, how- ever, was lower than growth in working capital. The scaling up of the orga- nized sector is reflected in the amount of inputs used. Between 2002 and 2012 the amount of fuel used in organized pulse processing mills went up, from 549 million rupees to 4,733 million rupees, a 700 percent increase over 10 years. Fixed capital includes both the plant and the machinery, among other things, and thus captures technology to the extent that it is embodied in the machinery. If the additional machinery is the same as that of the exist- ing machinery, it amounts to capacity addition. If the additional machinery is superior to the existing machinery, it is treated as technological progress (Bhavani, Gulati, and Roy 2006). Growth of fixed capital reflects capacity additions and/or technological progress and thus a rise in the potential scale of operation. Capital deepening can roughly represent technological progress, since new technologies are more capital-dependent and less labor-dependent. The variations in capital deepening evident across mills might be a function of an initial capital intensity that could vary by types of pulses as well as the final market. At the same time, on the supply side, factors like investment mea- sures and competition and policy regulations could determine the outcomes 5 New technologies vis-à-vis traditional manual technologies enable the production of hygienic and standardized products and can thus improve the level of sales. 148 Chapter 5 for pulse processing mills. There has been a substantial rise in the fixed investment in the organized pulse processing sector over time, from about 988 million rupees (in constant terms) in 2002 to 13,276 million rupees in 2012. This is due to a significant jump in the capacity— that is, the fixed assets per mill. This capacity expansion may be due to pent-up demand for most products that ignited market expectations for the industry, together with lib- eralization of investment restrictions on the supply side. On the employment side, it is possible that the quality of employment in the organized sector, in terms of labor productivity, would be much higher than that found in the unorganized sector. If so, it could be because labor in the organized sector has significantly more capital to work with. Geographic Distribution There is significant regional concentration in pulse processing by sector type across states (Figure 5.1). Four states account for the majority of orga- nized pulse processing. Strikingly, some big states (like Rajasthan and Uttar Pradesh) have very few organized sector mills. At the same time, Gujarat, although not a large producer of pulses, is comparatively industrialized and has a large number of organized mills. Highly agricultural states (like Punjab and Uttar Pradesh) do not contain a large number of mills in the orga- nized sector. Figure 5.2 presents the distribution of mills in the unorganized sector across states. It is notable that not only does a big state like Uttar Pradesh have a large number of unorganized sector mills (nearly 1,500), but a small state like TAbLE 5.2 Organized sector pulse processing over time, 2003–2012 (Rs millions at constant prices for values) Sl No. Item Unit 2003 2007 2012 1 number of factories numbers 900 1,315 1,535 2 fixed capital rs millions 1,294 2,184 13,276 3 Working capital rs millions 3,666 12,643 50,300 4 outstanding loans rs millions 3,142 11,287 37,437 5 Man days workers thousands 4,050 4,973 7,467 6 number of workers numbers 10,744 12,939 21,338 15 total inputs rs millions 36,949 93,216 478,268 17 Value of output rs millions 38,731 97,124 540,937 Source: authors’ calculations based on data from annual Survey of industries. StruCture of pulSe proCeSSing in india 149 Chhattisgarh (erstwhile part of Madhya Pradesh), which is a big pulse pro- ducer but is not industrialized, also has a large number of mills (1,161) in the unorganized sector compared to fewer than 100 mills in the organized sector. Also striking are the cases of undeveloped states, like Bihar (not shown) and Odisha (a state where pulse production has experienced a turnaround in recent times), where the organized processing sector has almost no presence. Figure 5.3 presents the employment shares of states in the organized and unorganized pulse processing sectors. Nearly 75 percent of employment in the organized sector is concentrated in the four states of Andhra Pradesh, FIGURE 5.1 Distribution of mills in the organized pulse processing sector, 2010– 2011 450 500 200 250 300 350 400 0 State Nu m be r o f u ni ts 50 100 150 Tr ip ur a Pu nj ab Od is ha De lh i As sa m Ka rn at ak a Ra ja st ha n Ut ta r P ra de sh Ta m il N ad u Ch ha tti sg ar h Gu jra t M ad hy a Pr ad es h M ah ar as ht ra An dh ra P ra de sh Source: data from annual Survey of industries. FIGURE 5.2 Distribution of mills in the unorganized pulse processing sector, 2010– 2011 600 800 1000 1200 1400 1600 1800 0 200 State Nu m be r o f u ni ts 400 Tr ip ur a Pu nj ab Jh ar kh an d As sa m De lh i Bi ha r Ha ry an a Ra ja st ha n M ad hy a Pr ad es h Or is sa Ka rn at ak a W es t B en ga l M ah ar as ht ra Ch at tis gr ah U tt ar P ra de sh A nd hr a Pr ad es h T am il N ad u Source: data from annual Survey of industries. 150 Chapter 5 Gujarat, Madhya Pradesh, and Maharashtra. In the unorganized sector, nearly 70 percent of employment is located in three states: Chhattisgarh, Tamil Nadu, and Uttar Pradesh. These findings are unsurprising given the similar distribution of mills across states in Figures 5.1 and 5.2, respectively. Figure 5.4 plots the distribution of mills in terms of their fixed capital across states. The pattern mirrors that of employment share, with fixed-capital shares FIGURE 5.3 Employment shares across states in the organized and unorganized pulse processing sectors in 2010– 2011 (%) Andhra Pradesh 26% Madhya Pradesh 11% Gujarat 11% Maharashtra 26% Organized sector Chhattisgarh 5% Others 21% Unorganized sector Chhattisgarh 41% Madhya Pradesh 3%West Bengal 4% Uttar Pradesh 22% Tamil Nadu 8% Orissa 5% Others 17% Source: data from annual Survey of industries and national Sample Survey organization (nSSo). FIGURE 5.4 Share of fixed capital across states in the organized and unorganized pulse processing sectors (%) Tamil Nadu 6% Others 12% Others 10% Uttar Pradesh 7% Andhra Pradesh 13% Maharashtra 38% Gujarat 12% Chhattisgarh 72% Madhya Pradesh 12% Madhya Pradesh 6% West Bengal 1% Uttar Pradesh 7% Tamil Nadu 4% Organized sector Unorganized sector Source: data from annual Survey of industries and national Sample Survey organization (nSSo). StruCture of pulSe proCeSSing in india 151 concentrated in a few states. Although pulse processing is spread across the country, these figures show concentration by type of industry. Alternatively, they imply that barring a few states, in a majority of states the processing remains dominated by the unorganized sector (in terms of the number of mills) with low technological intensity. The four dominant states in organized pulse processing account for nearly 80 percent of output in this sector. Productivity Growth: Shifts in Capital and Labor The growth in the organized sector of pulse processing could be happening at both the intensive and the extensive margins. New organized mills could be established but, equally, old (unorganized) mills could upgrade in terms of technology. There could be a drop in the number of unorganized mills at the same time as the number of organized mills is increasing, a Schumpeterian example of creative destruction. Over time, the states that have witnessed large drops in the number of unorganized pulse mills are Andhra Pradesh, Bihar, Madhya Pradesh, Rajasthan, and West Bengal. The growth in organized pulse processing has mainly been in urban units. The rural enterprises, meanwhile, remain dominated by unorganized mills. The share of capital per worker is higher in the leading states for organized pulse processing, so the spatial patterns are also reflected in labor productiv- ity. A few striking facts stand out regarding labor productvity. First, before 2009 output per worker was uniformly quite low across states, and after 2009 it switched to a significantly rising trajectory. It is possible the food price cri- sis of 2008 created pressure on the processing sector to become more efficient. At the same time, in the organized sector the variance in labor productivity across states also increased after 2009. In essence, capital deepening or tech- nological progress, which was comparatively uniform across states until 2007, became more variable after the food price crisis. Figure 5.5 and Figure 5.6 plot the output per factory and output per worker, respectively, in this sector over time. Meanwhile, in the unorganized sector, output per worker almost dou- bled over that decade, although it rose from a very low base. With greater productivity measured as output per worker, the wages and salaries are likely to have been higher in the organized pulse processing sec- tor after 2009 (Figure 5.6).6 There have been substantial increments in labor productivity in the organized sector, when measured this way, in frontline 6 The quality of employment, which determines the wages and hence incomes of employees, is as important as the quantity of employment generated. 152 Chapter 5 states after 2009. Empirical evidence for the changes in scale of operation of the pulses processing sector indicates that in the nine years following 2001, growth in the value of output per factory was quite modest. An average pulse processing factory in the organized sector produced annual output worth 85 million rupees at the beginning of the period and 300 million rupees at the end. At the factory level, growth in output may be due, on the supply side, to capacity additions or to technological progress, and on the demand side, it may be due to growth in the market. The supply-side factors enhance the capacity of a factory to produce more output, while the demand-side factors FIGURE 5.5 Output per mill in the organized pulse processing sector over time, 2001– 2012 350 400 250 300 150 200 0 2001 2002 Ou tp ut p er m ill (i n Rs m ill io n) 2003 2004 2005 2006 Year Output per mill 2007 2008 2009 2010 2011 2012 50 100 Source: authors’ calculations based on data from annual Survey of industries. FIGURE 5.6 Labor productivity across states over time in organized sector mills, 2002– 2012 Labor productivity across states Karnataka Chhattisgarh Andhra Pradesh Maharashtra Madhya Pradesh Gujarat Uttar Pradesh Tamil Nadu 60 0 10 20 30 40 50 Ou tp ut p er m ill (i n Rs m ill io n) 2002 2007 2010 2012 Year Source: data from annual Survey of industries. StruCture of pulSe proCeSSing in india 153 provide incentives to produce more. Consequently, variations in the growth of output per factory across mills could be due to variations in the supply and demand factors as well as to their interactions. Since we do not have data on market expansions for the mills, we try to capture capacity expansions and technological progress through the growth of fixed capital, capital deepening (the capital-to-labor ratio), and labor pro- ductivity (Figure 5.7). Expectations about the market growth and liberaliza- tion of investment regulations and the resulting forces of market competition might have prompted the organized-sector firms to bring in new technologies and add to their capacities. If so, that could be reflected in a rise in fixed cap- ital at the factory level. Capital deepening or capital intensity (expressed as capital-to-labor ratio) has been one of the conventional indicators of technol- ogy adoption. Determining the Production and Productivity of Organized Pulse Processing: A Regression Analysis Next, we conduct a rigorous analysis of the organized pulse processing sector using plant-level data. We are interested in assessing the role of state-specific characteristics and learning-by-doing factors in determining mill perfor- mance. In particular, we want to investigate the possible role of agglomeration, if any. Do mills that are older tend to be more productive? Does the presence FIGURE 5.7 Fixed capital per worker across states over time in organized sector mills, 2002– 2012 Fixed capital per worker Fi xe d ca pi ta l p er w or ke r ( in R s m ill io n) 2002 2007 2010 2012 Year 0 0.2 0.4 0.6 0.8 1.0 1.2 Chhattisgarh Madhya Pradesh Gujarat Maharashtra Andhra Pradesh Tamil Nadu Uttar Pradesh Source: data from annual Survey of industries. 154 Chapter 5 of more mills with larger output in a given neighborhood lead to better out- comes for the average mill in that neighborhood? These questions are impor- tant from a policy perspective. If agglomeration matters, then the promotion of new industry might require a critical mass of existing plants in the neigh- borhood. The answer could also be obtained by assessing the role of state fixed-effects in being associated with mill-level outcomes. Similarly, if experi- ence in the industry is important, then targeting firms or plants with a longer time in operation could be the optimal strategy. Basically, we are interested in estimating the following regression equation involving a repeated cross-section of mills in the organized sector: Oijt = αj + βt + θXijt + γnijt + δZijt + εijt (1) In equation 1, Oijt measures a specific outcome for mill i in state j at time t. Xijt is a matrix of characteristics of i in state j at time t. It excludes the two plant- and time-specific variables nijt and Zijt, respectively, that are entered as separate variables. nijt represents the number of years of operation of the mill i in state j at time t. Zijt is for the agglomeration term; it equals the total out- put or the number of mills in state j at time t excluding the ith mill. The larger this value is, the greater is the agglomeration factor with its network effect for the ith mill in state j at time t. A significant effect of this variable with the outcome variable would indicate that mills derive benefits from being in a locality that has more mills and/or that are producing larger output. nijt equals the year data recorded minus the year of inception. It captures the learning- by- doing effects where older plants are more productive (if there is a significant and positive coefficient). αj and βt respectively are the state and time fixed effects. The former captures whether, in relation to a benchmark state, the average outcome of a mill in a specific state is subpar or better. βt represents the time fixed effect, accounting for generic time factor that affects all mills across all states— for example, when the global food price crisis hit in 2008. These two fixed effects control for state-specific time-invariant unobserved characteris- tics, such as agroclimatic conditions if they are conducive to pulse cultivation. εijt is the classical error term. All standard errors are clustered at the state level. Apart from introducing state and time fixed effects separately, we also include in one specification state × time fixed effects. In the case of state fixed effects, we take Madhya Pradesh to be the bench- mark state, so state-specific effects are measured in relation to Madhya Pradesh. Several unobserved factors, such as governance, are not necessarily time-specific or state-specific but vary across states at different times. A state StruCture of pulSe proCeSSing in india 155 × time fixed effect accounts for such factors and minimizes omitted-variable bias. Examples of other variables included in Xijt are the capital-to-labor ratio employed in a specific mill in a state at a particular time, as well as mill- and time-specific use of inputs, mainly fuel and electricity. Because states vary in their degree of urbanization and therefore in the incidence of rural and urban location for the average mill, we also implement a specification with rural fixed effects separately. Results. Results from estimating equation 1 are presented in Table 5.3. From column 2 on, estimation results correspond to increasing levels of gen- erality. The first specification is the standard linear estimation using ordinary least squares. Subsequently, state and time fixed effects are introduced sepa- rately. One specification includes the rural location fixed effects to distinguish the outcomes between rural and urban mills. The coefficients on state fixed effects are presented separately. A few important points emerge from this estimation: • In the models the tendency for more experienced firms to be more produc- tive is validated. However, results for agglomeration effects are inconsis- tent, as one indicator is compatible with the hypothesis of agglomeration effects, while the other indicator points in the opposite direction. With the dependent variable as output per worker, the capital-to-labor ratio is (as expected) strongly associated with higher productivity. • The significant coefficient on electricity shows the important role it plays in affecting productivity in the pulse processing sector. Enterprise surveys in India often show that energy deficits are a binding constraint. Indeed, energy-deficient states such as Bihar and Uttar Pradesh have a negligi- ble-to-thin spread of the organized pulse processing industry. The bottom panel in Table 5.3 presents the coefficient of state fixed effects and rural fixed effects, respectively. Results show the following: • By taking Madhya Pradesh as the benchmark state, the labor productiv- ity in most states with a reasonable presence of pulse processing is subpar to varying degrees. Some frontline states, like Andhra Pradesh and Tamil Nadu, also perform badly in relation to Madhya Pradesh. • Moreover, there is a clear demarcation in the performance between rural and urban mills; in the organized sector, the average urban mill has signifi- cantly higher productivity than the average rural mill. 156 Chapter 5 TAbLE 5.3 Determinants of output per worker Explanatory variable Linear State fixed effect Rural-urban fixed effect Year fixed effect State x year fixed effect Capital-to-labor ratio 0.698*** 0.658*** 0.710*** 0.838*** 0.826*** (0.102) (0.1000) (0.100) (0.0953) (0.0973) petrol −0.0229 −0.0150 −0.0187 −0.0344 −0.0138 (0.0238) (0.0234) (0.0235) (0.0221) (0.0227) electricity 0.0173 0.0242** 0.0147 0.0241** 0.0164 (0.0121) (0.0121) (0.0119) (0.0112) (0.0129) number of years operational 0.000315*** 0.000385*** 0.000315*** −0.000380 0.000607** (5.02e−05) (5.11e−05) (4.95e−05) (0.000752) (0.000265) number of units in the state except this unit −0.0126*** −0.0156*** −0.0112*** −0.0105*** −0.0169** (0.00145) (0.00350) (0.00145) (0.00143) (0.00838) total output gen- erated by the state except this unit 9.27e−05*** 6.48e−05*** 8.69e−05*** 7.15e−05*** −2.39e−05 (1.01e−05) (1.21e−05) (9.96e−06) (1.03e−05) (5.11e−05) Constant 15.05*** 15.84*** 14.54*** 15.57*** 16.01*** (0.0746) (0.144) (0.103) (0.153) (0.226) observations 1,584 1,584 1,584 1,584 1,584 r-squared 0.184 0.244 0.209 0.304 0.409 Source: data from annual Survey of industries. Note: Standard errors in parentheses. *** p < 0.01, ** p < 0.05, * p < 0.1 State Fixed Effect—Base State: Madhya Pradesh himachal pradesh −1.928** (0.895) punjab −1.060** (0.464) uttaranchal −1.923** (0.896) haryana −3.253*** (0.267) rajasthan −0.694*** (0.233) (continued) StruCture of pulSe proCeSSing in india 157 For robustness, Tables 5A.1 and 5A.2 in the chapter appendix present the results of estimation with gross value-added per worker and output per mill as dependent variables. In all of these, our preferred specification is the one containing state × time fixed effects, which best minimizes the possibility of omitted variable bias, but identification of effects is difficult given the limited variation we are left with. State Fixed Effect—Base State: Madhya Pradesh uttar pradesh −0.499** (0.198) Bihar −2.786* (1.537) Manipur −5.137*** assam −2.350*** (0.783) orissa −2.062*** (0.641) Chhattisgarh −1.170*** (0.235) gujarat −1.101*** (0.177) Maharashtra −0.333* (0.181) andhra pradesh −0.659*** (0.233) Karnataka −0.904*** (0.234) Kerala −1.173 (1.538) tamil nadu −1.395*** (0.207) Rural-urban fixed effect urban 0.633*** (0.0907) Source: author’s calculations. Note: Standard errors in parentheses. *** p < 0.01, ** p < 0.05, * p < 0.1 TAbLE 5.3 (continued) 158 Chapter 5 Backward Links: Supply Chains to Pulse Processing Sufficient availability of pulses for the processing sector is a commonly dis- cussed problem (see, for example, the National Bank for Agriculture and Rural Development [NABARD n.d.]). Banerjee and Palke (2010) exam- ine the supply chain for pulses in general, and Yogan and Manohar (2015) examine the chain for chickpea. A common characteristic of the supply chains involving pulse processing is the large number of intermediaries that lie between the producers and the consumers. These intermediar- ies include commission agents, wholesalers, processors, and retailers. The marketing channels in pulses are both private and institutional. The insti- tutional arrangement for marketing includes procurement of pulses by pro- viding minimum support prices to the farmers through agencies like India’s National Agricultural Cooperative Marketing Federation (NAFED). Under this arrangement, the farmers sell to the procuring agency, and the millers source directly from the procuring agency. In the case of pulse processors, however, the amount of sourcing done through this channel is not signifi- cant. The private marketing channel is, in fact, the most common channel, and it exists for pulse processors throughout the country and for every type of pulse. Supply Chain Channels Four basic marketing channels for pulses, including processors, are identified by Banerjee and Palke (2010), which may be outlined as follows: • Channel 1. Farmer/producer → village trader → dal miller → wholesaler → retailer → consumer. • Channel 2. Producer → dal miller → retailer → consumer. • Channel 3. Producer → wholesaler → dal miller → retailer → consumer. • Channel 4. Farmer/producer → village trader → commission agent → dal miller → wholesaler → retailer → consumer. Channel 2, which links producers directly to millers, is comparatively rare. Similarly, millers getting pulses directly from wholesalers is also rare. The more common channels for dal mills are channels 1 and 4. In both these chan- nels the village traders are the initial link in the marketing chain. Traders can buy pulses directly from the farmgate and then supply them to processors, or else they can buy from the mandis (government-regulated wholesale markets) StruCture of pulSe proCeSSing in india 159 through a commission agent. Because there is very limited (if any) procure- ment by agencies like NAFED, farmers generally sell their pulses in their own villages, in the weekly markets, or in the nearby mandis. According to Banerjee and Palke (2010), farmers market about 75 percent of their pulse pro- duce and retain the rest for their own consumption and for seeds for the next year. In the case of chickpea, Yogan and Hansa (2015) look at the supply chain for the processing sector and find similar marketing channels as Banerjee and Palke (2010) do for the other pulses. Market Power, Price Formation, and Price Transmission in the Supply Chain In the pulses supply chain involving the processor, with limited or no pro- curement by the government or directly by the processors (channel 2), most farmers sell to traders. The policy-driven entry barriers in trading mean that traders enjoy certain market power, particularly in relation to the small farmers. For example, to begin business operations to market pulses, any purchaser/dealer/trader needs to take two licenses (India, Ministry of Agriculture 2012): 1. License under the respective state Agricultural Produce Marketing Committee (APMC) Act to deal in agricultural produce (please see below for an explanation about the APMC Act). 2. License to stock pulses under the Essential Commodity Act— Pulses Control Order. Furthermore, in some states, the traders/commission agents seeking a license are required to have a physical establishment for such business in the APMC market area. These provisions mean that there are comparatively few trad- ers in agricultural commodities in general and pulses in particular, giving traders some degree of monopsony power. Farmers who are located away from the mandis (the wholesale markets) usually sell their produce to traders at the farmgate where the farmer’s bargaining power is even weaker. The few trad- ers who pick up pulses from the farmgate usually discount the price to include transportation cost to the mandi. The problem is compounded because of asymmetric information. Located away from the market, these farmers lack information about prices and are not able to bargain for the best price. The information consists of data and analysis containing inventory, facility, trans- portation, price, and customers as well (Yogan and Manohar 2015). In the 160 Chapter 5 pulses supply chain, including the processors, the buyers and sellers (small farmers) lack information about the external market price given the limited coverage of channel 2. With numerous intermediaries, flow of information and market signals are manipulated that minimize the returns to the farmers and affect the supply to the processors. Hence, in the mandis as well as at the farmgate, one of the results of imper- fections is suppression of prices accruing to the farmers. Chapter 3 shows an implicit collusion among traders around the focal point of market support price (MSP). Also, imperfections in the market result in limited transmission of market prices to the farmgate level (see Rahman 2015 for asymmetric price transmission in the case of pulses). Because of low farmgate prices relative to retail prices, we also see small supply responses to rises in pulse prices because farmers receive only a small fraction (less than 50 percent) of what buyers such as processors pay in the market. Aggregation of the small surpluses through producer companies can possi- bly help in this context. For example, a recent case study in Tamil Nadu shows that farmers’ realization increased from 47 percent to 63 percent of the retail price of white lentils once the growers organized themselves into a producer company (Angles and Karunakaran 2016). A number of farmer producer organizations (FPOs) have been organized for pulse growers across differ- ent parts of India, but there is a large variation in their performance. We need more research to understand how to promote successful and viable FPOs that bring more benefits to their members and also help the processors. Successful FPOs will not only help in the marketing of pulses but may also act as effec- tive channels of extension to promote the use of better seeds, lifesaving irri- gation, and best practices in pulse production and thereby ensure a consistent supply of good quality raw materials to the mills. One other change related to marketing that can bring benefits to farmers and millers alike could be to free pulses from APMC taxes. Under the APMC Act, all transactions are regulated to take place in government-licensed whole- sale markets (mandis). The state governments then impose taxes on all trans- actions that take place. The buyers have to pay these taxes and they can build it into their bids, which result in price markups. Mandis in different states have different taxes. Some states, like Haryana and Uttar Pradesh, have very high taxes (15 percent and 19 percent, respectively). In the major pulse-pro- ducing state of Madhya Pradesh, the taxes are as high as 9 percent. These taxes add to increases in consumer prices and reduced farmer prices in pulses, and they should, therefore, be done away with. StruCture of pulSe proCeSSing in india 161 Additional Costs through the Supply Chain The marketing costs in the supply chain leading to the processing sector nor- mally include (1) handling charges at local points, (2) assembling charges, (3) transport and storage costs, and (4) handling charges by wholesalers and retailers. In addition, a market fee is charged either on the basis of weight or on the basis of the value of the produce and is usually collected from the buy- ers. The seller or the buyer (or sometimes both) pays this commission to the commission agents. On top of handling and marketing fees, across India there are numerous taxes to be paid, such as a toll tax, terminal tax, sales tax, and octroi.7 All these taxes vary across the markets and from state to state, and the rates are different as well. Usually, the taxes are payable by the seller, but they are built into the prices the processors pay for the raw material. Miscellaneous charges to cover handling, weighing, loading, unloading, and cleaning are pay- able either by the seller or by the buyer. Due to this complex set of levies— market fees, commissions, taxes, han- dling and transport charges, and other miscellaneous charges— the absolute value of the total marketing margin varies widely from market to market, from channel to channel, and from one period to another. For example, in the Azadpur mandi in Delhi, officially the commission is 6 percent, but in practice it goes up to 10 percent. In the Vashi market in Navi Mumbai, the officially notified commission is 8 percent, but in practice it runs as high as 15 percent (Banerjee and Palke 2010). Both these markets deal in different types of pulses as well. Based on a field survey, Banerjee and Palke (2010) iden- tified the marketing costs for two types of supply chain involving processors of pigeon pea. They computed the marketing cost and margins for the two most important channels (channel 1 and channel 4) from the producer all the way through the processor (reproduced in Table 5.4). As Table 5.4 shows, moving from channel 1 to channel 4 adds costs in terms of the mandi tax and cess as well as the commission for the agents.8 These marketing costs are based on sourcing from within the same state where the processor is located. If raw material is brought in from outside the state, other taxes and charges will be applied, compressing the margins further and augmenting the cost of raw materials for the processors. 7 Octroi is a tax levied on goods entering a town or city. In India, it is generally imposed by large cities. 8 The government imposes cess (for example, education cess). 162 Chapter 5 Constraints in Processing The major problems for the vast majority of India’s present-day pulse process- ing units are their low product recovery rates and their high milling costs, all stemming from the fact that these processing units are still running on the old, traditional system rather than deploying such modern, sophisticated sys- tems as those used in Australia, Canada, Germany, and Spain (see Banerjee TAbLE 5.4 Marketing costs–supply chain for pulse processor Channel 1 Percent to the next links purchase price Channel 4 Particulars Rupees per quintal Particulars Rupees per quintal 1 producers’ sale price/village traders’purchase price 2,000 producers’ sale price/village trad- ers’ purchase price 2,000 2 Cost incurred by producer/farmer Cost incurred by producer/farmer a Cost of gunny bags 25 1.14 Cost of gunny bags 25 B loading 3 0.14 loading 3 C unloading, weigh- ing, and cleaning 8 0.36 unloading, weigh- ing, and cleaning 8 d transportation 39 1.77 transportation 39 total cost (a+B+C+d) 75 3.41 total cost (a+B+C+d) 75 3 Village traders’ margin 125 5.68 Village traders’ margin 125 4 Village traders’ selling price 2,200 100 Village traders’ selling price 2,200 5 processors’ pur- chase price 2,200 63.77 processors’ pur- chase price apMC tax and cess 55 agents’ commission 44 6 fixed operational cost of dal mill 738.5 21.41 fixed operational cost of dal mill 738.55 total cost 837.55 7 processors’ margin 511.45 14.82 processors’ margin 511.45 8 processors’ selling price 3,450 100 processors’ selling price 3,549 Source: Banerjee and palke (2010). Note: apMC tax and cess are imposed by government-regulated wholesale markets. StruCture of pulSe proCeSSing in india 163 and Palke 2010). In the pigeon pea sample of Banerjee and Palke (2010), almost all the mills were found to be running according to the traditional sys- tem, which cause higher milling losses in the form of fragmentation and pow- der, resulting in a lower recovery of dal than the modern methods. In addition, the average capacity utilization of these processing units in Banerjee and Palke’s sample was just 70 percent, due to the recurring nonavail- ability of raw pulses and the way processing was operated as a seasonal activity. The units used batch processing, which involves excessive material handling that in turn results in pulse loss. They would prepare a lot of 50 to 60 quin- tals of pretreated/conditioned pulses at a time for milling, and only after that batch had been converted into dal was the process repeated. Moreover, most of the observed units used sun drying, which reduces their capacity utilization during the rainy season. Moreover, because pulses are aggregated from a large number of players (from either channel), they differ in their quality, variety, and size, but grading and standard-setting for pulses is lacking among the pro- cessors. This is because pulses are all mixed together and this can compromise quality. Also, processors are made to sort quality, which imposes costs in terms of time and other resources. In some states, an additional constraint is the lack of an uninterrupted supply of electricity or an uninterrupted supply of water (or both). These are major concerns. Another constraint is access to financing. The establishment of a dal mill unit involves a large investment in block and working capital. In Banerjee and Palke’s sample, the working capital, which constituted around 85 percent to 90 percent of their overall cost of operations, was the most important com- ponent. The units obtained loans from informal sources—with interest rates as high as 15 to 20 percent per year—to procure pulses (Banerjee and Palke 2010). Stocking limits represent another constraint. These are limits the government imposes on processors to check speculation and hoarding, and they are often binding. These limits vary enormously from state to state and across time. As of March 2015, for example, Bihar was placing a stocking limit of 1,500 quintals on all pulse mills in the state. In Gujarat, the limit for unmilled pulses was 500 quintals, and 250 quintals for the finished stock of milled pulse. Haryana had a limit of 2,000 quintals. Karnataka put in a regu- lation requiring that stocks not exceed 30 days’ requirement. In Maharashtra the limit for unmilled pulses was equal to one-ninth of a mill’s annual produc- tion/installed capacity, and for milled pulses it was one-eighteenth of annual production/installed capacity. Other states had varying levels of stock limits. The stocking adversely affects the functioning of pulse processors (Yogan and Manohar 2015). 164 Chapter 5 High price fluctuations in pulses are another constraint, as they discourage processors from stocking larger inventories even within the limit. The govern- ment’s MSP, which serves only as a benchmark price for pulses (since there is no or limited procurement), has been quite variable over the years. For 2010– 2011, 2011– 2012, 2012– 2013, and 2014– 2015, the MSP for chickpea has moved from 2,100 to 2,800, to 3,000, and finally to 3,100 rupees per quintal. With this degree of price fluctuation in raw material prices, staying financially sustainable in the industry becomes a challenge, especially for small-scale mill- ers and manufacturers. There are additional constraints that discourage pro- cessors from purchasing outside their own area or state. When millers are located in the same area where the mandi is situated, they have the advan- tage of buying raw materials directly from the wholesale market. But millers located outside a producing state cannot buy products directly; because their inventory is normally limited to a maximum of 10 days’ consumption, they do not make bulk purchases partly because there are stocking limits. It is unlikely to be economical to travel to make direct purchases of small quantities, so they must depend instead on agents. As discussed earlier, transacting across state borders also brings in additional costs and charges. Forward Links between Processor and Consumer Banerjee and Palke (2010) present data on how the consumer price is arrived at from the processors to the consumers. Going forward from the proces- sors to the consumer, it is either through the wholesalers to the retailers or to retailers directly who then sell to consumers. Banerjee and Palke show that in this chain, processing costs comprise 67 percent to 71 percent of the differ- ence between the farmgate price and the consumer price. In other words, in the chain a sizable portion is contributed by the processing costs. Hence, in the formal or quasi-formal supply chains, significant reduction in consumer prices can be achieved through improved efficiency in pulses processing. As discussed, the pulses supply chain generally has the feature that just when con- sumer prices are high, the producer prices tend to be low (Yogan and Manohar 2015). Based on Banerjee and Palke (2010) primary data, farmgate prices tend to be on average half of the consumer price. They are a bit higher in channel 4 over channel 2. The lack of direct link between processors and retail is salient here. When the processor sells forward, it includes marketing costs compris- ing cost of labor, weighing, cleaning, packaging, transportation, and pro- cessor’s margin. Banerjee and Palke (2010) estimate the wholesale and retail margins to be about 2 percent each in the forward link between processor and StruCture of pulSe proCeSSing in india 165 consumer. Bantilan et al. (2014) estimate large effects of improved technical efficiency in chickpea processing on consumer prices. Hence supply chain effi- ciencies through improved processing and better production technology for growers can likely bring down consumer prices of pulses. Policies for Improving Pulse Processing Policies need to be designed to address several issues confronting the pulse processing sector. Addressing the supply problem. The uninterrupted supply of raw mate- rial is a prerequisite for running pulse processing units efficiently. Yogan and Manohar (2015) and Banerjee and Palke (2010) both show that this has been a binding factor for pulse processing. Apart from resorting to imports, pro- cessing units cannot maintain continuous operations all year round because the available domestic production has generally been inadequate. Efforts should, therefore, be made to ensure a supply of raw material through- out the year. Supply chains need to be developed adequately. In the case of Tamil Nadu’s processing mills, for example, chickpeas must be supplied from Andhra Pradesh, Karnataka, Madhya Pradesh, and Maharashtra. In addition, imported raw material is purchased from both private parties and from gov- ernment institutions. Relying on multiple sources has been the only solution for several processing units to meet their requirements for raw material. In this context, systems that would encourage direct purchase by the processors from farmers could be helpful— that is, after diluting APMC restrictions. Direct marketing would enable farmers as well as processors to economize on trans- portation costs and also improve price realization. Because pulses are sourced from a large number of small farmers, there is a need to establish a system of grading and sorting to standardize the inputs. Addressing farmers’ marketing challenges. Along with the buyers, farmers face such problems as delayed payments and lack of bargaining power as com- pared with licensed traders as well as a lack of adequate infrastructure in the mandis from where the processors source their pulses. Even in states that have diluted or done away with the APMC Act, new markets have not replaced them, forcing most farmers to sell their produce to traders who, as aggregators, supply to different buyers, including processors. Addressing inefficiencies in mill performance. There are avenues for exploiting new technologies in pulse processing, some of which have been developed in India itself, as identified in studies like those by NABARD and Tamil Nadu Agricultural University (TNAU). For example, institutes like 166 Chapter 5 Central Food Technological Research Institute (CFTRI) in Mysore have developed a conditioning technique to loosen the husk without resorting to sun drying and oil and water application. This step has been mechanized with the introduction of conditioning units. Use of this conditioning tech- nique, as developed by CFTRI-Mysore, could be one option for improving mill performance. The present rate of losses can be greatly minimized through the use of improved dal mills. Several research institutes in India have developed improved dal mill technologies that are highly versatile and energy efficient.9 The improved dal mills have a dehusking efficiency of about 95 percent, and their yield of split pulses is about 80 percent to 85 percent (Banerjee and Palke 2010). Greater use of these technologies could be helpful for the growth and sustainability of pulse processing. Addressing marketing needs of the processors. Finally, in marketing, unlike other branded products such as basmati rice and edible oils, apart from besan no branded product of any pulse is currently popular (Banerjee and Palke 2010). Processed pulses have seen a significant spread of branded prod- ucts in recent times, with processing companies like Halidram and Bikano, but in dal the branding and product differentiation has so far been limited. How to improve branding and product differentiation in dal is an important problem to address. Consolidation of Firms and Organizing of Farmers The problems that the processing sector has faced due to both irregular sup- ply of raw materials and inadequate quality (leading to excess capacity) can be mitigated by better coordination with the farmers (Banerjee and Palke 2010). The analysis in this chapter shows that there is a gradual process of consol- idation happening on the firm side, with the processing sector undergoing scaling up and concentration of firms. The growth in sales of processed food has also led to an increase in processing firms’ use of pulses. The path to firm consolidation in processing is likely to follow three stages (Bora, Gulati, and Roy 2006). In the initial stage, small processors dominate, but with increas- ing income and urbanization, changes in food habits, and the entry of for- eign direct investment (FDI), firms can scale up their operations. New firms 9 These include PDKV in Akola, CFTRI in Mysore, Gobind Ballav Pant University of Agriculture and Technology (GBPUAT) in Pantnagar, CIAE in Bhopal, Indian Institute of Pulse Research (IIPR) in Kanpur, Tamil Nadu Agricultural University (TNAU), and Coimbatore and Indian Agriculture Research Institute (IARI) in New Delhi. StruCture of pulSe proCeSSing in india 167 come in with product differentiation, and a constant process of churning is observed. However, by the third and final stage only the most efficient firms— those that can adapt to the demands of the market— survive, leading to a high level of concentration with a few scaled-up firms dominating the market. In India, the empirical evidence shows that there has been a definite scaling up in the food-processing sector (Bhavani, Gulati, and Roy 2006). As this chapter has shown, the levels of output and capital per firm have gone up substantially in pulse processing since 2001. It seems that pulse processing is currently in stage 2. With the right policy support, a transition to stage 3 can happen, and if it is managed well, it can be an opportunity to make pulse pro- cessing dynamic. In dealing with the farmers, who are mostly smallholders, creating scale is important. Pulse growers are increasingly organizing themselves into farmer producer organizations (FPOs), especially in the states that are lead- ing producers, such as Madhya Pradesh and Maharashtra. Indeed, the gov- ernment’s integrated program for the development of 60,000 pulse villages, which is managed by the Ministry of Agriculture through its Small Farmers Agribusiness Consortium (SFAC), explicitly lists the promotion of FPOs in the pulse sector as a policy to improve outcomes. The SFAC argues that to work out economies of scale and link pulse farmers to markets, it is necessary to organize pulse farmers into groups, both for getting access to quality inputs and for creating market links, such as those with processors. The government announced 2014 as “the year of FPOs” and several FPOs in pulses were initi- ated (India, Ministry of Agriculture 2013). To date, though, there have been only a few cases of contract farming by the pulse processing sector. Examples include the Odisha Rural Development and Marketing Society (ORMAS) for pigeon pea. Under this arrangement, which is managed by self-help groups, the processor has an assured market by supplying pulses for food programs like the Mid-Day Meal Scheme. This ini- tiative has resulted in ensuring a good income to the members of the self-help groups. In 2010, ORMAS procured about 6,000 quintals of local pigeon pea through these groups (Sharma 2010). A private-sector example of contract farming in pulses processing is that of the Tata Chemicals Company (through its subsidiary, Rallis India), which markets differentiated pulses after process- ing by keeping them unpolished. The product is sold with its brand of pulses, called i-Shakti. Tata Chemicals purchases pulses largely from the Wardha Cotton and Soya Producer Company, which has a mill in Karnataka and recently opened one in Maharashtra. 168 Chapter 5 There is also the case of an NGO that began contracting for pulse pro- cessing in 2009. The NGO, known as Seva Mandir, is based in Udaipur in Rajasthan and its aim is to make farmers independent of the intermediaries for selling their produce. The arrangement comprises a cluster of seven vil- lages where the production of pigeon pea is significant. The processing mill is run by a farmer producers’ group consisting of local pulse farmers, while Seva Mandir has played a major role in marketing the produce. Currently, the size of the farmer group is small, comprising 86 farmers, but access to the mill is not limited to members. Other pulses procured for processing are green gram and black matpe, and the NGO uses prices in neighboring wholesale markets as reference prices. Like the Tata i-Shakti example, the Seva Mandir processing arrangement is one based on introducing product differentiation and brand- ing. Both arrangements maintain that the pulses they market must be free of chemical fertilizers and pesticides, and both marketed them as premium organic produce. It is notable that in the Gulbarga district of Karnataka, the “pigeon pea bowl” of India, where there are nearly 300 mills and aggregate production can be high, there are very few instances of contract farming. Instead, most pro- cessors buy from government-regulated wholesale markets. Karnataka state government has taken a step forward, nevertheless, by introducing e-tender- ing, and it has developed a common market across the state that can help the processors meet the scale requirement and get good-quality raw material (see Chengappa et al. 2012 and Athawale 2014 for details). As pulse processing has come up in the adjoining states of Andhra Pradesh and Maharashtra (with a large number of organized sector mills), the Gulbarga processing sector has shrunk a bit. Conclusion This chapter presented an analysis of pulse processing using secondary-data sources for 2002– 2012 for both the organized and the unorganized sectors. Several key messages emerge from the analysis. First, although the unorga- nized sector dominates in pulse processing in terms of the number of mills and employment, the organized sector dominates in terms of sales value. The analysis shows that a shift is taking place, away from the unorganized sec- tor and moving toward the dominance of the organized sector. Significant restructuring is clearly taking place, with the proportion of large mills increas- ing, bringing a corresponding increase in installed capacity. This shift, StruCture of pulSe proCeSSing in india 169 however, is more pronounced in some states than in others, likely because of learning effects, in addition to the individual state’s supporting infrastruc- ture. State-specific factors seem to be quite important in determining the sec- tor’s outcomes. The pulse processing sector does not seem to be a big generator of employment, either in the organized or the unorganized sector. The major employment generation that does occur is located at the back end— that is, in pulse production, which gets a strong boost from the existence of process- ing links. For these links, the quality of inputs and their consistent supply remain as concerns. The level of technology used in pulse processing needs to be raised, particularly in the unorganized sector. When compared with other countries, India’s pulse processing sector has a low capital-to-labor ratio and only moderate capital deepening over time. In many Southeast Asian coun- tries, small-scale rural food-processing industries have functioned as growth engines (Sharma, Panthania, and Vashist 2003). Right now, the demand for processed pulses in India remains primarily for items that require only pri- mary processing, implying that value-addition in pulses through secondary processing has been limited. Indian consumers are highly price-sensitive, so a reduction in the cost of processed pulses through more efficient processing is needed to raise demand and consumption. Currently, there is low capacity utilization, and that in turn leads to higher processing costs. Although we do not have data on the exact capacity utilization in pulse processing, based on the standard in the Indian food-processing sector as a whole, we can estimate that it is less than 70 percent. Part of the reason for the lack of capacity utilization is struc- tural, with the seasonal nature of the activity itself leaving mills idle except when imports or stored inputs are channeled to run the mills uninterrupted. There might be limits to scaling up the pulse processing sector. Before it can become an engine of growth, it must undergo a structural shift toward larger mills with higher productivity and with institutional arrangements that can ensure regular supply of good quality pulses. According to Banerjee and Palke (2010), to add value to the pulses, the processing units need to be equipped so as to meet consumer demand. They recommend improving the supply of good-quality raw materials, adopting modern conditioning techniques to loosen the husk without resorting to sun drying, and extending support for storage facility construction and other infrastructure. The few recently introduced models of Tata i-Shakti, on a larger scale, and of Seva Mandir, on a smaller scale, may offer insight into ways the processing and marketing of pulses as health food may help realize the potential of pulse processing. With the changes under way in the Indian food system, what is 170 Chapter 5 needed is a new way of doing business, a new approach based on innovative institutions that can cut transaction and marketing costs for both firms and farms. There is a need to scale up success stories like that of the Indian dairy cooperative movement, where the processing sector played a pivotal role. As suggested by the chief economic adviser to the government of India, there might be a need to apply the Amul model for pulses.10 The solutions also require that government play a complementary role in the building of appropriate infrastructure and institutional support. Institutional bottlenecks require significant policy changes to induce effi- ciency. For example, the taxation rate on processed foods in India is one of the world’s highest. Dev and Rao (2004) state that the net tax level is 21– 23 percent on food items in India, while the comparative tax burden is 10 percent in the Philippines, Indonesia, and Malaysia; 14– 15 percent in the Netherlands and the UK; and 17 percent in China and Ireland. It is also nec- essary that a uniform value-added tax is imposed on all states to facilitate growth. The current move toward a Generalized System of Taxes (GST) is a big policy step that is likely to boost the pulse processing sector. Similarly, the recent announcement regarding a National Agricultural Market could help processors source materials from a larger area. Several states have started amending the APMC Act, which is crucial for market reform. Toward that end, India’s Ministry of Agriculture had formulated a model law on agricul- tural marketing in consultation with state governments, which enables the establishment of private markets/yards, direct purchase centers, consumers/ farmers markets for direct sale, and promotion of public-private partner- ships (PPPs) in the management and development of agricultural markets. The regulation and promotion of contract farming arrangements are part of this legislation. References ASI (Annual Survey of Industries). 2001– 2015. India, Ministry of Statistics and Programme Implementation. Central Statistics Office. Industrial Statistics Wing. www.csoisw.gov.in /cms/En/1023-annual-survey-of-industries.aspx. Athawale, G. 2014. “APMC and E-trading for Financial Inclusiveness in Karnataka.” IBMRD Journal of Management and Research 3 (2): 84– 98. 10 The Amul model involves firm-farm links where the state-run dairy cooperatives link with dairy processing. This system has been largely credited with making India the world’s largest producer of milk. See “Amul Model” 2015. StruCture of pulSe proCeSSing in india 171 http://www.csoisw.gov.in/cms/En/1023-annual-survey-of-industries.aspx http://www.csoisw.gov.in/cms/En/1023-annual-survey-of-industries.aspx “‘Amul Model’ Needed for Pulses: Arvind Subramanian.” 2015. Accessed February 2016. www.thehindu.com/news/national/other-states/amul-model-needed-for-pulses-arvind -subramanian/article7904394.ece. Angles, S., and K. R. Karunakaran. 2016. “Value-Chain Development for Blackgram in Tamil Nadu through Group Marketing.” Paper presented at the Conference on Pulses for Sustainable Agriculture and Human Health, New Delhi, May 31– June 1. Banerjee, G., and L. M. Palke. 2010. “An Overview of Pulses.” In Economics of Pulses Production and Processing in India, 1– 18. Department of Economic Analysis and Research, National Bank for Agriculture and Rural Development (NABARD), Occasional Paper 51. Bantilan, M. C. S., D. Kumara Charyulu, P. Gaur, D. Moses Shyam, and J. S. Davis. 2014. “Short Duration Chickpea Technology: Enabling Legumes Revolution in Andhra Pradesh, India.” Research Report 23, Patancheru, India, ICRISAT. www.icrisat.org/ what-we-do/mip/SPIA .pdf. Accessed September 2016. Bhavani, T. A., A. Gulati, and D. Roy. 2006. “Structure of the Indian Food Processing Industry: Have Reforms Made a Difference?” In Plate to Plough: Agricultural Diversification and Its Implications for the Smallholders in India. Submitted to Ford Foundation. Washington, DC: International Food Policy Research Institute (IFPRI). Bora S., A. Gulati, and D. Roy. 2006. “Fragmenting Bottom and Consolidating Top: India’s Changing Food System and Implications for Small Holders.” Mimeo. International Food Policy Research Institute (IFPRI), Washington, DC. Chavan, J. K., D. M. Shere, H. K. Jawale, and D. K. Salunkhe. 1983. “Effects of Soak Treatment to Legume Seeds on the Cooking Quality of Resultant Dal.” Indian Journal of Nutrition and Dietetics 20: 249– 254. Chengappa, P. G., M. Arun, C. G. Yadava, and H. M. Prasanna Kumar. 2012. “IT Application in Agricultural Marketing Service Delivery— Electronic Tender System in Regulated Markets.” Agricultural Economics Research Review 25: 359– 372. Dev, M. S., and N. C. Rao. 2004. “Food Processing in Andhra Pradesh Opportunities and Challenges.” Working Paper 57. Centre for Economic and Social Studies, Begumpet, Hyderabad. Geervani, P., and F. Theophilus. 1980. “Effect of Home Processing on the Protein Quality of Selected Legumes.” Journal of Food Science 42: 1269– 1275. Gulati, A., P. K. Joshi, and M. Landes. 2008: “Contract Farming in India: An Introduction.” Accessed June 2016. www.ncap.res.in/contract_%20farming/Resources/1.Introduction.pdf. ICRISAT (International Crops Research Institute for the Semi-Arid Tropics). 1991. Uses of Tropical Grain Legumes. Proceedings of a Consultants Meeting, March 27– 30. Accessed September 2016. http://oar .icrisat.org/964/1/RA_000177.pdf. 172 Chapter 5 http://www.thehindu.com/news/national/other-states/amul-model-needed-for-pulses-arvind-subramanian/article7904394.ece http://www.thehindu.com/news/national/other-states/amul-model-needed-for-pulses-arvind-subramanian/article7904394.ece http://www.icrisat.org/ what-we-do/mip/SPIA.pdf http://www.icrisat.org/ what-we-do/mip/SPIA.pdf http://www.ncap.res.in/contract_%20farming/Resources/1.Introduction.pdf http://oar.icrisat.org/964/1/RA_000177.pdf India, DSIR (Department of Scientific and Industrial Research). 2007. “Food Processing— Indian Regulatory and Policy Environment.” Accessed June 2016. www.dsir.nic.in/reports/isr1 /Food%20Processing/6_5.pdf. —. 2012. “Expert Group on Pulses (Alagh Committee Report), 2012.” Ministry of Agriculture, Department of Agriculture and Cooperation (DAC), New Delhi. —. 2013. Policy and Process Guidelines for Farmer Producer Organizations. Ministry of Agriculture, Department of Agriculture and Cooperation (DAC), New Delhi. Accessed June 2016. http://nhm.nic.in/Archive/FPO-Policy&Process -GuidelinesDAC2013.pdf. Jambunathan, R., S. D. Hall, P. Sudhir, V. Rajan, and V. Sadhana, 1991. Uses of Tropical Grain Legumes. Proceedings of a Consultants Meeting, March 27– 30, 1989. International Crops Research Institute for the Semi-Arid Tropics. NAAS (National Academy of Agricultural Sciences). 2006. “Employment Opportunities in Farm and Non-Farm Sectors Through Technological Interventions with Emphasis on Primary Value Addition.” Policy paper 37. Accessed July 2016. http://naasindia.org/Policy%20Papers /policy%2037.pdf. NABARD (National Bank for Agriculture and Rural Development). N.d. Knowledge Banks. Dal Mills. Accessed July 2016. https://www.nabard.org/english/dal.aspx. Parpia, H. A. B. 1973. “Utilization Problems in Food Legumes.” In Nutritional Improvement of Food Legumes by Breeding, 281– 295. Proceedings of a symposium, July 3– 5, 1972, Rome, Italy. New York, NY, US: Protein Advisory Group. Rahman, A. 2015. “Is Price Transmission in the Indian Pulses Market Asymmetric?” Journal of Quantitative Economics 13 (1): 129– 146. Rama Rao, G. 1974. “Effect of Heat on the Proteins of Groundnut and Bengal Gram (Cicer arietinum).” Indian Journal of Nutrition and Dietetics 11: 268– 273. Sharma, K. D., M. S. Pathania, and G. D. Vashist. 2003. “Role of Rural Women in Small Scale Agro-Processing Sector: An Economic Analysis of Samridhi Mahila Processing Co-operative Society in Himachal Pradesh.” Indian Journal of Agricultural Economics 58 (3): 578– 588. Sharma, P. 2010. A Study on Agricultural Marketing System in Odisha. Jaipur, Rajasthan, India: National Institute of Agricultural Marketing (NIAM). TNAU (Tamil Nadu Agriculture University) Portal. 2016. “Post Harvest Technology: Agriculture. Pulses.” Accessed September 2016. www.agritech.tnau.ac.in/postharvest/pht_pulses.html. Vijaya Kumar, M., ed. 2006. “Employment Opportunities in Farm and Non-Farm Sectors through Technological Interventions with Emphasis on Primary Value Addition.” Policy paper 37. National Academy of Agricultural Sciences (NAAS), New Delhi, October. StruCture of pulSe proCeSSing in india 173 http://www.dsir.nic.in/reports/isr1/Food%20Processing/6_5.pdf http://www.dsir.nic.in/reports/isr1/Food%20Processing/6_5.pdf http://nhm.nic.in/Archive/FPO-Policy&Process-GuidelinesDAC2013.pdf http://naasindia.org/Policy%20Papers/policy%2037.pdf http://naasindia.org/Policy%20Papers/policy%2037.pdf https://www.nabard.org/english/dal.aspx http://www.agritech.tnau.ac.in/postharvest/pht_pulses.html Yogan, V., and M. L. Hansa. 2015. “Supply Chain Management of Pulses (Chickpeas) Industry in India.” International Journal of Economic Research 12 (2): 319– 329. Yogan, V., and H. L. Manohar. 2015. “Supply Chain Management of Pulses (Chickpeas in India).” International Journal of Economics Research 12 (2): 319– 329. Appendix TAbLE 5A.1 Determinants of gross value-added per worker Explanatory variable Linear State fixed effect Rural urban fixed effect Year fixed effect State cross year fixed effect Capital-to-labor ratio 0.956*** 0.936*** 0.964*** 0.924*** 0.899*** (0.0716) (0.0706) (0.0714) (0.0718) (0.0756) petrol −0.0180 −0.0171 −0.0164 −0.0156 −0.00369 (0.0157) (0.0154) (0.0157) (0.0157) (0.0163) electricity 0.0269*** 0.0280*** 0.0259*** 0.0254*** 0.0224** (0.00795) (0.00805) (0.00793) (0.00793) (0.00947) number of years since operational 0.000261*** 0.000315*** 0.000260*** −0.000424 0.000346* (3.33e−05) (3.41e−05) (3.32e−05) (0.000511) (0.000188) number of units in the state except this unit −0.00934*** −0.0108*** −0.00880*** −0.00800*** −0.0172*** (0.00101) (0.00233) (0.00102) (0.00107) (0.00593) total output produced by state except this unit 6.13e−05*** 3.53e−05*** 5.90e−05*** 5.22e−05*** 3.08e−05 (6.74e−06) (8.37e−06) (6.76e−06) (7.27e−06) (3.70e−05) Constant 12.14*** 12.60*** 11.97*** 12.29*** 12.65*** (0.0508) (0.0966) (0.0723) (0.107) (0.166) observations 1,326 1,326 1,326 1,326 1,326 r-squared 0.346 0.392 0.351 0.358 0.438 174 Chapter 5 State Fixed Effect: Base State—Madhya Pradesh Jammu and Kashmir −2.595*** (0.969) punjab −0.741** (0.320) uttaranchal −2.424*** (0.566) haryana −0.941* (0.485) rajasthan −0.526*** (0.161) uttar pradesh −0.510*** (0.133) Manipur −2.201** (0.969) assam −1.543*** (0.495) Jharkhand 1.457** (0.566) orissa −1.439*** (0.492) Chhattisgarh −0.760*** (0.166) gujarat −0.514*** (0.120) andhra pradesh −0.422*** (0.154) Karnataka −0.448*** (0.160) tamil nadu −0.640*** (0.136) urban 0.207*** (0.0638) Source: authors’ estimations. StruCture of pulSe proCeSSing in india 175 TAbLE 5A.2 Determinants of output per mill Explanatory variable Linear State fixed effect Rural–urban fixed effect Year fixed effect State cross- year fixed effect Capital-to-labor ratio 0.563*** 0.531*** 0.571*** 0.710*** 0.725*** (0.111) (0.109) (0.110) (0.104) (0.106) petrol 0.0242 0.0350 0.0272 0.0114 0.0436* (0.0259) (0.0254) (0.0258) (0.0242) (0.0246) electricity 0.0286** 0.0427*** 0.0267** 0.0359*** 0.0265* (0.0132) (0.0131) (0.0131) (0.0123) (0.0140) number of years operational 0.000257*** 0.000314*** 0.000256*** −0.000317 0.000836*** (5.46e−05) (5.54e−05) (5.42e−05) (0.000823) (0.000288) number of units in the state except this unit −0.0153*** −0.0154*** −0.0143*** −0.0135*** −0.0235** (0.00158) (0.00380) (0.00159) (0.00157) (0.00910) total output produced by the state except this unit 0.000106*** 7.89e−05*** 0.000102*** 8.47e−05*** −6.81e−05 (1.09e−05) (1.32e−05) (1.09e−05) (1.13e−05) (5.55e−05) Constant 17.25*** 17.76*** 16.89*** 17.78*** 18.25*** (0.0811) (0.157) (0.113) (0.167) (0.246) observations 1,584 1,584 1,584 1,584 1,584 r-squared 0.197 0.258 0.207 0.305 0.419 176 Chapter 5 State Fixed Effect: Base State—Madhya Pradesh himachal pradesh −2.291** (0.972) uttaranchal −1.984** (0.973) haryana −4.816*** (0.711) delhi −0.200 (0.290) rajasthan −0.442* (0.253) Bihar −3.607** (1.669) Manipur −5.959*** (1.669) assam −2.484*** orissa −2.335*** (0.696) Chhattisgarh −0.920*** (0.256) gujarat −0.583*** (0.193) andhra pradesh −0.676*** (0.253) Karnataka −0.498* (0.254) Kerala −0.000906 (1.669) tamil nadu −1.044*** (0.225) Rural-urban fixed effect _iru_Code_2 0.454*** (0.0994) Source: authors’ calculations. Note: Standard errors in parentheses. *** p < 0.01, ** p < 0.05, * p < 0.1 StruCture of pulSe proCeSSing in india 177