Evolution and prospects of the rice mill industry in Uganda

For the development of rice production in sub-Saharan Africa, it is critical to improve the quality of domestic rice produced and sold in the market. Being at the entry point in the upstream of the postharvest marketing chains, rice mills hold the pivotal position in the improvement of rice quality. Based on data collected from rice mills all over Uganda, this paper presents the evolution of the rice mill industry in the last two decades. It looks into the present state of rice mill operation, examines the structure of the rice milling market, analyzes the cost and returns structure of rice mill operation, and discusses the future directions of the rice mill industry. The rice mill industry in Uganda has developed at a remarkable pace since the early 2000s. The high profitability of rice milling business and a low entry barrier facilitated small scale rice mills to have grown rapidly. Rapid increase of the number of rice mills still continues particularly in newly emerging rice growing areas but the growth begins to stagnate in traditional rice growing regions. The majority of rice mills are small-sized but some large-scale ones have been emerging. In traditional rice growing regions, the rate of milling fee is declining due to increasing competition and begins to reflect the difference in rice quality. Under the scale economy inherent in rice mill operation and growing quality consciousness in the rice milling market, milling machines would become larger and more sophisticated replacing the current competitive milling market with oligopolistic one. It is important to develop efficient small to medium scale milling system with emphasis on quality improvement features such as de-stoning and grading so that small scale rice mills can be competitive step by step countervailing against the scale economy of large scale rice mills.


INTRODUCTION
Rice in sub-Saharan Africa (SSA) has recently been attracting profound attention of researchers, policy makers and international aid agencies as a crop of strategic importance for food security and poverty reduction (Nwanze et al., 2006;Somado et al., 2008;CARD, 2008;Campbell et al., 2009;Seck et al., 2010Seck et al., , 2012;;Africa Rice, 2011;EUCORD, 2012;Demont, 2013;Demont et al., 2013).Rice cultivation has a relatively longer history in West Africa than in East Africa where it is not a traditional crop (Balasubramanian et al., 2007).
In West and East Africa alike, however, the demand for rice has been increasing in the recent decades most rapidly among major food sources (Seck et al., 2013).Although rice production in SSA has been increasing, it does not match the increase in demand.As a result, nearly 40% of rice consumed in SSA is imported from outside the region.The skyrocketing price of imported rice during 2007 to 2008 which induced food riots in many countries in SSA, revealed how vulnerable food security was in the region (Moseley et al., 2010;Seck et al., 2012Seck et al., , 2013;;Demont, 2013).
All the studies cited above points to the need of increasing domestic rice production.Indeed, long before the food crisis of 2007 to 2008, it has been a standing policy target of SSA countries to increase domestic production so as to reduce the dependence on rice imports on account of enhancing food security, saving foreign exchange for economic development and improving rural livelihood (Jayne and Jones, 1997;Diagana et al., 1999;Moseley et al., 2010;Demont, 2013).Among various reasons that have been put forward for SSA countries not being able to attain this policy target, the related literature is unanimous in pointing out the quality of rice produced domestically in the region.The quality of domestic rice in SSA is said to be inferior to the extent that it does not have any substitutability for imported rice.
The poor quality of domestic rice is primarily due to the poor postharvest handling of rice especially at the milling stage that determines the quality (Lançon et al., 2004;Becker and Yoboué, 2009;Campbell et al., 2009;Moseley et al., 2010;Demont and Neven, 2013;Futakuchi et al., 2013;Stryker, 2013).Being at the entry point in the upstream of the postharvest marketing chains, rice mills hold the pivotal position in the improvement of the quality of rice in the market.There are many instances on farm that need improvements for higher rice quality (Futakuchi et al., 2013), but even if all of them are remedied, it is obvious that the rice quality problems persist as long as the quality of rice milling remains poor.Naturally, almost all the studies, reports and documents cited above claim that the present state of affairs at rice mills in SSA countries be changed so as to improve the quality of milling.Beyond that common understanding, however, the literature is largely silent as to the state and characteristics of the rice mill industry in SSA. 1his paper tries to fill this gap by providing information as to the present situation of the rice mill industry in Uganda.As in other countries in East Africa, rice is not a traditional staple in Uganda but its consumption began to increase rapidly at around 1990,followed by its domestic production that has especially accelerated since the early 2000s when the government launched and promoted New Rice for Africa (NERICA) as upland rice (Kikuchi et al., 2014).Although the rice production of the country increased at a remarkable rate of more than 10% per annum from 1990 to 2012, 30% of the consumption was satisfied by import in 2012 (Kikuchi et al., 2013).The prime target of the rice policy newly set by the government was to reduce rice import (MAAIF, 2009).
Uganda also shares with other SSA countries the same structure of rice consumption market that is clearly dichotomized into two segments by quality; the higher and lower quality markets.In the case of Uganda, the lower segment is formed by public markets, handling the majority of rice consumed in the country of ordinary quality with a very limited number of 'brands' sold at reasonable prices friendly to the majority of consumers in the country.The other segment is formed by supermarkets with a minor quantity but high quality of rice, mostly imported one, sold under a large number of brands, the price of which is wide open towards a very high end (Kikuchi et.al., 2013).These two segments have begun overlapping each other at the margin.As the economy develops resulting in an increase in household income coupled with population growth, the rapid increase in the demand for rice will continue.For domestic rice production to meet demand, it is critical to improve its quality so as to be qualified for the upper segment.The need for improving the quality of domestic rice through upgrading the quality of milling is what stakeholders in the rice industry unanimously point out (Wilfred 2006, Emerging Market Group, 2008;PMA Secretariat, 2009;MAAIF, 2009;Oyee, 2009;Chemonics International, 2010;SMJR Consult, 2012).However, studies on the rice mill industry have been few in the country with an exception of Candia et al. (2008).In this respect, the situation in Uganda is representative of many SSA countries.2Based on data collected from rice mills, this paper addresses the following objectives: i) to present the evolution of the rice mill industry in the last two decades, ii) to look into current state of rice mills and their operations, iii) to examine the structure of the rice milling market, iv) to analyze the cost and return structure of rice mill operation, and iv) to envisage future directions of the rice mill industry.In this paper, the term rice mill, rice miller and rice mill owner mean a management body operating a rice mill plant with one or more rice milling machines and are used interchangeably.

Rice mill surveys
Data from a series of rice mill surveys were used in this study.In To obtain data on rice mill operation, we interviewed rice millers in 25 districts in the four regions in two phases, March to June and July to September in 2012.Except for large-scale rice mills, our sample rice millers were drawn from the population in two stages; we first selected sample districts and then chose sample rice millers for interview.The selection of sample districts was purposive to represent major rice producing areas in the four regions.The sampling of the second stage was not random in the strict sense of statistics textbooks but sample millers were selected randomly in the sample districts whenever we came across rice mills and their owners or operators were available for interview.The number of sample rice millers is shown in Table 1 by region.The sampling ratio for the country as a whole was 11%.Although the 70 sample rice millers include 4 large scale rice mills, which were distinguished from small rice mills in terms of the type and capacity of rice milling machine used as well as the scale 3 For the estimation details, see Kikuchi et al. (2013).
of operation, this study mostly concerns with small rice mills, which represent the overwhelming majority (99%) of rice mills in the country.
Slightly different questionnaires were used in the two phases: basic information about mills, such as the year of establishment, the number, type and capacity of milling machines, workshop and storage, milling fee rate, variety, source and quantity of rice milled, and disposal of bran, was common in both phases; questions about the rice prices at mills and the destinations of milled rice after milling were specific to the first phase questionnaire; and data on the costs of mill operation, including maintenance costs, to the second phase.
To attain the stated objectives, we first present the evolution and current state of operation of the rice mill industry, summarizing the data collected from the surveys in descriptive tables.Among important parameters that demonstrate the state of rice mill operation, the determinants of milling capacity and rice mill maintenance cost are examined by means of regression analysis.Second, we examine the state and structure of the rice milling market by looking into the level of rice milling fee among the regions.
Third, we estimate the cost and return structure of rice mill operation by estimating the fixed and variable costs.Fourth, we analyze the break-even scale of rice mill operation by means of the revenue-cost ratio.
The revenue-cost ratio of rice mill operation is computed by dividing its revenue by its costs.The revenue of rice mill operation is defined in this study as follows: R = () Q, Where R = revenue of rice mil (Ush), Q = quantity of paddy rice milled (kg), = rate of milling fee (Ush/kg),  = milled rice recovery rate (%), = price of bran (Ush/kg),  = rate of bran produced per kg of paddy (%), and  = share of bran taken by rice mill (%).The cost function of rice mill operation is defined as follows: Where C = total cost of rice mill operation (Ush), Q = quantity of paddy rice milled (kg), FC = fixed cost (Ush), and VC(*) = variable cost (Ush).Fixed cost is defined as consisting of the capital depreciation of milling machine, the rent for workshop and storage space and manager's salary, if any.The variable cost is defined as the summation of the electricity bill for machine operation, the cost of collecting paddy from villages, labor wages and maintenance costs, all of which are a function of the quantity of rice milled.Since the machine depreciation, the rate of electricity consumption and maintenance costs vary according to the type and the capacity of the machine (HP), the cost function is defined by type and capacity of milling machine.The revenue-cost ratio is computed as R/C.

Evolution of the rice mill industry
The history of rice cultivation in Uganda can be traced as far back as to the latter half of the 19 th century when it was grown for a handful of Arab and Swahili traders but its cultivation of a significant extent began in the early 1970s.The production has been accelerated in the 1990s through the 2000s alongside with rapid increases in rice import triggered by low prices in the world rice market.Among the regions in the country, the Eastern region has been the forerunner of the present rice boom followed by the Northern region.Rice cultivation in the Western region began in the early 2000s with the introduction of NERICA and within less than 10 years the region has become a major rice producing area.The Central region shares a similar pattern in rice production as the West, though at a lesser extent.The evolution of rice mill industry has mirrored the evolution of rice production (the top chart of Figure 1).The rice mill with the oldest history among the sample rice mills had been operating since 1983, and a few of them followed in the early 1990s.These rice mills of old establishment were all found in the East, indicating that the initial spurt of the rice mill industry had begun in the late 1990s in the same region, prior to the start of the promotion of upland rice in the early 2000s.
In the North and the West, some rice mills established in the late 1990s.Rice mills established in the late-1990s in the West were first established as maize mills and started rice milling operation in the 2000s in addition to maize milling.The start of the NERICA promotion in the early 2000s accelerated the increase in the number of rice mills: as much as one-third of the rice mills operating at present were established in this period.
The regional distribution and the rapid growth of the rice mill industry in recent years are confirmed for the total number of rice mills (Table 1).It is estimated that there were 645 rice mills in the country in 2012, of which more than 50% concentrated in the East.The total number of rice mills increased by 50% in 5 years from 2007 to 2012.
The rate of growth was particularly high in the North and West where upland cultivation with NERICA varieties had been vigorously promoted since the early 2000s.The 'rice boom' in cultivation has thus accompanied the 'rice mill boom'.It should be noted however, that the increase in the number of rice mills had decelerated in most recent years.The North and West, the latter in particular, still showed very high rates of increase for 2009 to 2012 but the rate of change for the East and Central for this period was substantially lower than that for 2007 to 2009.Rice mills can be classified according to the type of rice milling machine they use.Rice milling machines adopted by rice millers in Uganda can be broadly grouped into three types: Engelberg (including improved types with forced air), mill-top (one-pass) and large scale machines with grader, de-stoner, etc., in the order of machine sophistication (Candia et al., 2008).The bottom chart of Figure 1 shows the number of mills by type of machine. 5n early years, all rice mills used Engelberg. 6However, the number of rice mills with more sophisticated mill-top machines, first appeared in the late 1990s, has been increasing faster than the number of those with Engelberg.Among the sample rice millers, the first one who installed a large scale milling machine appeared in the mid-2000s and the number has increased to four since then.As will be explained later, there are some differences between Engelberg and mill-top machines in terms of the capacity and quality of rice milling and the acquisition prices but the differences in these respects are substantially larger between these two 'conventional' machines and the modern large scale machine. 7We call the rice mills with the 'conventional' machines small rice mill as against those with large scale machine, though there are some differences in the scale of operation among the small rice millers. 8he rice milling business has attracted people from a wide spectrum of occupations.Many rice millers were initially in maize milling business: about 20% of the sample rice millers used to be maize millers until they started rice milling as the demand for it increased.The farm-related sectors were also major pools that produced rice millers: about 40% of rice millers were farmers and crop traders.This indicates that the sudden rise of the demand for rice milling has given a business chance to entrepreneurial farmers in rural areas.The rest of 40% of rice millers were from various business circles, some related to farming, such as seed production and farm-product processing, but many totally outside the agriculture sector, such as banker, bus owner and car repair shop.Some professionals, such as government civil servants, teachers and medical doctors, also joined the rice mill industry.All this suggests that rice milling is a lucrative business.
It also suggests that the rice mill industry is easy to enter, financially as well as technically.Since the 'conventional' milling machines often become out-oforder, the knowledge on mechanics helps the rice mill operation, but generally no special knowledge and skill are required.As far as the 'conventional' machines are concerned, the capital requirement for establishing a rice mill, just to install a rice mill machine (or machines) in a small workshop with small storage space, is affordable.These characteristics of the industry combined reduce the entry barrier to the rice mill industry.The relatively low requirement of initial investments is reflected on the fact that the majority of the sample rice millers could finance their initial investments for establishing their mills out of own savings.It is worth noting that the process of rice mill evolution in Uganda has largely been a result of the autonomous efforts made by entrepreneurs in the private sector to cope with the suddenly surging demand for rice milling services nearly without any government interventions such as subsidies.

Rice mill plant
The list of capital goods that a rice miller has to provide to operate a rice mill is very simple: rice milling machine, mill workshop and storage space.

Rice milling machines
The most important capital equipment for rice millers is rice milling machine.Table 2 shows that the 70 sample rice millers own 98 rice milling machines; 1.4 machines /miller on average.It was found that China dominates as the country supplying rice milling machines to Uganda.For the power source of milling machines, 70% are electricity operated, and the rest were diesel operated or electricity operated with a stand-by diesel engine.Rice millers preferred electricity-operated machine to dieseloperated one for lower maintenance and energy costs, but frequent power failures that stop the mill operation often make it necessary to use diesel-operated machines.
Most of the milling machines used by rice millers were of horse power (HP) from 20HP to 30HP: 60% of the machines fell in this range of HP.The full range of HP for the two major types of machines operated by small rice millers was from15HP to 50HP.However, most of Engelberg machines (Models N70, N120, etc.) were found in the HP classes of 20-30HP with the mode at 24-25HP, while mill-top machines (Models SB 10, SB 50, etc.) distributed more evenly over the six HP classes with the mode at 30HP.The distribution of milling machines by type has clear regional patterns.Engelberg machines were found mostly in the East with history of longer rice cultivation, while mill-top machines were found mostly in newly emerging upland rice growing areas in the West, Central and North.

Rice mill workshop and storage space
Other capital facilities necessary for rice millers are workshop where rice milling machines are installed and storage space or warehouse where they store sacks of milled rice after milling.For small rice mills, on average, a rice mill operated in a workshop of 33m 2 with additional storage capacity of 42m 2 , though they varied widely.The workshops of many small rice millers were not only small but shabby buildings or humble cottages.

Rice mill operations
With the milling machines and plants explained above, rice millers operate their mills to produce milling services.This section presents the operations of small rice mills.

Quantity of rice milled
On average, the quantity of paddy rice milled by the sample small rice mills in the year 2011-12 was 350 t/mill (Table 3).The variation around this average was large as shown in Figure 2 which reveals a bi-modal distribution of rice mills, bifurcating the sample rice millers into two groups at the average scale of operation; about 60% of sample rice mills belonged to the small-scale operation group and 40% to the medium-scale operation group.The bi-modal pattern is more distinct for rice mills with Engelberg machines.For rice mills with mill-top machines, the concentration in the scale classes from below 50 t to 200 -300 t was more than for Ebgelberg, but at the same time, there were more rice mills whose machines milled more than 600 t/year.Such a bi-modal pattern in the scale distribution of rice mills appears to suggest that there is the minimum scale of operation in rice milling which differs according to the capacity of milling machines and the type of machine used.The quantity of rice milled being one of critical determinants of the profitability of rice milling operation, it is vitally important for rice millers to secure sufficient amounts of paddy rice to mill all year around, and rice mills make various efforts for this end.The second part of Table 3 shows how rice mills acquire rice for milling.About 70 and 20% of paddy rice for milling was brought to the mills by farmers and village collectors/ middlemen, respectively, and the rest of about 10% was collected by mills by going to rice growing villages to buy paddy rice from farmers or sending their agents to the villages for the acquisition.It is indicative that the direct acquisition of paddy from farmers was particularly important for rice millers in the East, where the rice mill intensity was highest.Judging from the average cost of collecting paddy rice from villages (Ush 4000/sack of 100 kg), rice mills tried to collect paddy rice from villages on average in the radius of around 70 km (Kikuchi et al. 2013).Rice millers' efforts to collect as much rice for milling as possible brought them to villages not only within the same district but in other districts or even in neighboring countries (the third part of Table 3).For all, about threequarters of paddy rice milled was from within the district where the mill operated and the rest was brought from outside the district.It is suggested that the rice milling market was relatively more competitive in the East and the North, so that rice millers there tried to obtain rice by visiting villages in and outside the district they operate (the fourth part of Table 3).In contrast, in the West, more than 90% of rice milled by millers is brought to the mills from within the district.Together with the observation that the percentage of rice collected by millers' own efforts was low in the West, the rice milling market in this region was relatively less competitive than the East.Being at the initial node along the post-harvest rice marketing chains, rice millers were generally operating as rice traders as well (the fifth part of Table 3).The West was an exception in this respect; the percentage of 'pure' rice millers who specialized only to rice milling service was as high as 60%.
Aside from going out to distant villages to acquire paddy rice, rice millers operating in the competitive rice milling market make various efforts to secure as much rice for milling as possible, for example, to pay farmers or middlemen the whole or a part of the cost for transporting paddy rice from villages to the mill, and to lend money to farmers, particularly in planting and weeding seasons when farmers need cash for hiring labor on condition that rice is brought to the mill when harvested (Sakurai et al., 2006).

Milling recovery and milling capacity
In addition to the quantity of paddy rice to mill, there are two technical parameters that affect the output and performance of rice milling business: milled-rice-recovery rate and milling capacity of rice milling machines.The milled-rice-recovery rate is defined as the ratio of the quantity of milled rice after milling to the quantity of paddy rice before milling.
On average for sample rice mills, this rate was 65%, which tallies with the average over African countries reported in IRRI (2013).More than 70% of rice mills fell in the range of 60 to 70% and no statistically significant difference in the recovery rate was found among the regions.The milling capacity was an indicator to measure the efficiency of rice milling machines, defined as the quantity of rice that a milling machine can mill within a unit of time.This efficiency would depend on machine's HP, power source, machine type, vintage of machine, etc.The number of milling machines by HP and by power source is shown in the third part of Table 2.For electricityoperated machines, 30 HP was the most popular horsepower.The average milling capacity of machines is shown in parenthesis after the HP class.As expected, the milling capacity increases as the machine's HP becomes larger.
Further to examine the determinants of milling capacity, we conducted regression analyses for the milling machines of 50HP or less, by regressing machines' milling capacity on several explanatory variables.The results are summarized in Table 4 as Regression #I and #II.In addition to machines' HP, Engelberg dummy gives a coefficient statistically significant at the 5% level, indicating that, ceteris paribus, Engelberg machines are less efficient than mill-top machines.All other variables were not significant.When we estimate the costs and returns of rice milling operation later, the milling capacity estimated by Regression #II shall be used.

Mill workers
For running rice mill operation, rice millers need workforce consisting of managers, machine operators, operator assistants and some others.An average rice mill employs one manager, 4 milling machine operators and assistants (2 workers in a team; work in 8 to12 hour shift in the busy seasons) and 4 other workers (drying /cleaning rice, carrying rice sacks, etc.).The mode of salary /wage payment for these employees depends on the job type; a time rate for manager, piece rate for other workers and time or piece rate for machine operators.
The time rate payment is usually in monthly salary for managers and monthly or daily wage for operators.The piece rate wage is paid either in cash or in kind per sack of milled rice worked.

Maintenance
All rice mills practice the maintenance of rice milling machines, such as greasing and changing spare-parts (bearings, screen, rubber roll, etc.) at certain intervals.However, the variation in their maintenance practices was large.On average, rice millers with Engelberg machine practiced maintenance every 39 days (s.d.=50; N=13) with the expenditure of Ush 905,000 (s.d.=925,000) and those with mill-top machine every 62 days (s.d.=49; N=10) with Ush 607,000 (s.d.=563,000).Regression analyses, conducted to examine factors that affected the maintenance expenditure, revealed that it was a function of quantity of rice milled, in the double-log form (Regression #III and #IV in Table 4).The coefficient of the logarithm of quantity milled was positive and highly significant.Moreover, the coefficient was not statistically different from unity, indicating that the maintenance expenditures increase just proportionally to the increase in the quantity of rice milled.Though at the 10% significance level, the milling capacity gave a positive, significant coefficient.The coefficient of Engelberg dummy was not significant.The coefficient of maintenance interval is positive, indicating the longer the interval the larger the maintenance expenditure, but not statistically significant at conventional significance levels.Since the retaining this variable in the regression improves the adjusted R 2 substantially, Regression #IV is going to be used to predict the maintenance expenditure. 9

Milling fee
Rice millers receive milling fee as a return to their milling service.Without exception, the fee is in a piece rate per kg of rice milled.Milling fee rates adopted by the sample rice millers are summarized in Table 5 by region and by type of machine.The average fee for the entire sample was Ush 141 /kg of milled rice.Since the average price of milled rice at rice mill was Ush 3000 /kg for the most 9 Since maintenance expenditure and quantity milled or milling capacity could be both cause and affect each other, The estimation by the ordinary least squares (OLS) may suffer from the endogeneity problem.We use the OLS results assuming that the observational errors in our data outweigh the biases, if any, resulted from the endogeneity.
popular rice variety of Supa,10 the milling cost accounted for 4.7%, which was lower than the rate of 6.7% reported by Futakuchi and Sakurai (2006) for a land-locked rice growing area in Ghana.Unlike rice price, which are fairly uniform for a certain variety group not only within each region but also across regions (Kikuchi et al., 2013), milling fee differs significantly within each region as well as among regions.Moreover, it differs among the types of milling machine.
The within-region variations in the fee rate may be due to the location-bound nature of rice milling service that easily creates a locally isolated monopolistic market.More interesting are the across-region variations that could be resulted from the difference in the degree of competition in the rice milling market among regions.For all, milling fee was lowest in the East where more than 50% of rice mills in the country were concentrated (Table 1), highest in the Central where rice cultivation had begun most recently and in-between in the North and the West.Though the fee rate in the West was not statistically different from that in other regions.In the East, in particular, the fee adopted by rice mills with Engelberg machines was on average Ush 107 and there were rice millers who reduced it to a level as low as Ush 70.All this suggests that the rice milling market in the East was so competitive that many rice millers were compelled to reduce their milling fee.
It is also interesting to observe the variation in milling fee among different types of machine since it could be brought about by the difference in the quality of milling (Sakurai et al., 2006).Among the three types of milling machine, Engelberg is ranked the poorest in terms of milling quality.Although mill-top performs better than Engelberg (Sakurai et al., 2006), the degree of quality difference is far more significant between mill-top and large-scale machines (Candia et al., 2008).Table 5 shows that for the average over all the regions, the milling fee for Engelberg was significantly lower than that for mill-top and this result was maintained even if the regions were controlled.Such a result could be an indication that the difference in milling quality was reflected in milling fee in the rice milling market.The Eastern rice growing region offers an interesting case where the average milling fee for large-scale rice mills was significantly higher than for Engelberg (P=0.002) but significantly lower than for mill-top (P=0.005).This may suggest that the large-scale rice mills intentionally set their milling fee so as to be in-between, though the small number of observations for both mill-top and large-scale rice mills hinders us from being conclusive.

By-product
Another source of revenue for rice millers is rice bran that rice milling process generates at a rate about 15% of rice milled.The bran is sold as animal feeds or as a material for organic fertilizer.Buyers of bran are not only local pig and poultry farmers but also feed traders in various places including Kampala and Kenya.Since the price of rice bran at rice mill was, on average, around Ush 80/kg, its value took around 10% of the milling fee.Who takes rice bran after milling differs among regions.For instance, in the West, no sample rice millers took rice bran, leaving its disposal to farmers.On average for the entire samples, the bran was taken by rice mills and farmers equally.

Costs and returns of mill operation
The estimation of the cost-return structure was made for rice millers who operate with one milling machine, assuming two typical cases with respect to machine, scale of operation and milling fee: (A) Engelberg 20HP (electricity-operated) with 200 t/year of paddy rice milled adopting the milling fee of Ush 120/kg and (B) Mill-top 50HP (electricity-operated) with 600 t /year of paddy rice milled adopting the milling fee of Ush 160/kg.The basic data used in the estimation are summarized in Table 6 with some details of cost estimation in the footnotes, and the results of estimation are presented in Table 7, which elicits the basic cost-return structure of rice milling operation.
For 'small-scale' operation, the largest cost item was electricity bill that accounts for 41% of the total cost, while for 'medium-scale' operation, labor costs took the largest share of as much as 40%, followed by electricity.For both cases, maintenance costs took only a small share, 2 and 6%, respectively.The share of fixed costs was 20% for Engelberg and 22% for mill-top, the majority of which was accounted for by the rent for workshop and storage for both cases.The depreciation cost of the milling machine was also a small component of the total cost, 3% and 5%, respectively, which demonstrates the low entry barrier of the milling industry.
In the revenue side, milling fee accounted for about Tokida et al. 2569 95% of the total revenue for both cases.Deducting the total cost from the total revenue, 28% of the revenue was left with the mill owner as surplus for the Engelberg case and 39% for the Mill-top case.The rate of returns to the initial investment on the milling machine was estimated 164% /year for the Engelberg case and 256% for the Milltop case.On average, rice milling is a very lucrative business.Such high profitability should have been the main factor that brought about the 'rice mill boom' in recent years.

Break-even scale of operation
However, the profitability of rice mill operation is highly sensitive to the quantity of rice milled and the milling fee rate.For example, if the quantity milled in Case A in Table 7 is reduced to 50 t, with other assumptions unchanged, mill owners' surplus dives down to a negative figure.
Likewise, if milling fee in Case B is reduced to Ush 100 /kg, the rate of return drops down to 21%.
In search of the break-even scale of rice mill operation in the rice milling market where the milling fee rate is subject to downward pressure because of increasing competition among rice mills, the revenue-cost ratio is computed for four milling machine cases and for different levels of milling fee, the results of which are depicted in Figure 3.The milling fee rates of Ush 100, Ush 150 and Ush 200 are popular rates found in various regions for both Engelberg and mill-top machines.The milling fee of Ush 70/kg is the minimum rate actually adopted by some rice millers in the East.For the most popular milling fee rate of Ush 100 as of 2012, the break-even scale is 80 to 100 t/year for small HP machines and 300 to 400 t/year for large HP machines, regardless of the type of machine.Any scale more than these gives a positive surplus to mill owners.For the fee rate of Ush 150 and above, the break-even scale is reduced to 40 t/year or less and 150 t/year or less, respectively.These results are consistent with the bi-modal distribution of rice mill operation observed in Figure 2: the small-and medium-scale operation groups being demarcated by the average operation scale of 350 t/year.If the milling fee rate is lowered to Ush 70 as observed in the East, the quantity of rice that rice millers have to mill to clear the break-even scale increases to 550 t/year for Engelberg 20HP and 260 t/year for mill-top 20HP, and far more than 1000 t/year for large HP machines for both types.
These results suggest that if the competition in the rice milling market becomes so harsh that the fee rate is reduced to such a low level, single-machine small rice millers whose scale of operation is less than 200 to 300 t/year might have to exit the rice milling industry giving their market shares to rice millers of relatively larger scale of operation.
Another way for rice millers to deal with the lower fee 1) For small rice millers.2) Estimated using the milling capacity function (Regression #II in Table 4).3) Average of sample rice millers.4) Typical usable life.The capital depreciation by the linear depreciation over the usable lifetime.5) Estimated by type of machine.6) From Table 3. 7) Assume typical case for each machine type.8) Average acquisition price converted to 2012 prices by the GDP deflator.9) Ush 487.6/KWh + tax (18%).The rate of power charge is of 'intermediate', among the charges for peak, intermediate and off-peak.10) From Table 5. 11) Average cost that millers collect paddy rice from villages, obtained from sample millers.12) Typical wage rates adopted by sample rice millers.13) Average scrap value = Ush 500/kg, obtained from sample rice millers.14) Assumed from the first and second modes in Figure 2, respectively.15) Estimated using the maintenance function (Regression #IV in Table 4).16) Quantity milled (in milled rice) / milling capacity + time for idling, adjustment etc. (20% of then milling time).levels may be to increase the number of rice milling machines, as some rice millers are actually doing.An approximate break-even scale per machine for a doublemachine rice mill can be estimated by cutting the fixed costs per machine into half, which gives the break-even scale per machine that is about half of the single-machine mill; under the milling fee of Ush 70/kg, 270 t for Engelberg 20HP and 125 t for mill-top 20HP.
This makes double-machine millers' break-even scale per mill 540 t and 250 t, respectively; not much improvements from the single-machine mill as far as the break-even scale is concerned.For Engelberg machine, however, it could be possible for double-machine millers to improve the quality of rice milling by using one unit for removing husks and the second one for whitening.The serial use of the machines could also reduce heat damages and broken grains by lowering the pressure in the milling chamber.If this technical change in the rice milling process could improve the quality of milling to the level mil-top machines attain and if, as a result, the double-machine Engelberg millers could enjoy a milling fee level that is higher than the poor quality fee level by about 30%, as is the case in Table 5, their break-even scale would be close to the case of 'Fee=100' in Figure 3, which illustrates the importance of technical changes in milling process toward higher quality rice milling. 1) For small rice millers.Data are from Table 6.2) Cost items with (FC) are fixed cost.3)Capital depreciation of milling machine, obtained after deducting the salvage value of the machine from the machine price.4)For the investment on milling machine.

Conclusions: Future direction of the industry
The rice mill industry in Uganda has developed at a remarkable pace since the early 2000s.The high profitability of rice milling business and a low entry barrier have attracted entrepreneurs from various social circles.
In newly emerging rice growing regions, the industry is still developing rapidly, but in the East with relatively longer history of rice mill development, the industry is entering a mature stage.The emergence of rice milling fee lower than Ush 100/kg in the East, in contrast to higher fee levels of Ush 150/kg to Ush 200/kg prevailing in newly emerging rice growing regions, is an apparent sign of the degree of competition that the rice milling market is undergoing, and in this respect, the maturing stage is nothing but the stage of adjustments.
Considering the rapid pace of the development, it is a matter of time that the intensified competition brings the rice milling market in the newly emerging rice growing regions into this stage.
The scale economy inherent in the rice mill operation inevitably makes the changes that occur in this adjustment stage a shift of rice mills towards larger scale of operation.The emergence of large scale rice mills in the East and the North is a genuine sign of the direction that the rice mill industry is heading.For small rice mills, the changes that would take place are shifts from 'old fashioned' Engelberg to mill-top machines, or even more modern machines, and from small HP to large HP machines or single machine to multi-machine operations.Although the quality of rice is not so important yet in the national rice market, price differentiation due to quality difference, cleanliness in particular, has been emerging.
Accordingly, the milling fee rate has begun to be differentiated according to the quality of rice milling.The demand for rice of higher quality will increase steadily as the Ugandan economy grows, accelerating the need for the rice mill industry to be more and more quality conscious.Such market-led changes also work as a pressure for small rice mills with a small HP machine to exit the industry, strengthening the polarization of the industry into small rice mills with small scale of operation and ones with large scale of operation.Whereas the former satisfies the demand for milling services from customers who do not care much the quality of rice, the latter, including large scale rice mills, aims at the higher segment of the rice consumer market.As all large scale rice mills are equipped with pre-cleaners, de-stoners, separators and graders, there is a trend among small rice mills to introduce these facilities.In order for the rice mill industry to be a socially sound and economically viable industry, it is highly desirable to develop these facilities as technologies as divisible as possible, so that small rice mills can improve their rice milling quality step by step countervailing the scale economy of the large scale rice mills.

Figure 1 .
Figure 1.Cumulative number of sample rice mills by year of establishment.

4 4
In Uganda, NERICA 4, the most popular upland rice variety at present, was officially released in 2002 .
small rice millers.2) Brought by farmers.3) Brought by village collectors / middlemen and town traders.4) Acquired from farmers by mills by going and / or sending agents to villages.5) Acquired paddy for milling from within the district where the mill was situated.6) Acquired paddy from other districts, including foreign countries (Kenya and DRC).7) Not including the own district.

Figure 2 .
Figure 2. Number of rice mills by quantity of paddy rice milled per mill per year and by type of machine, 2012.

Figure 3 .
Figure 3. Revenue-cost ratio of rice mill operation by type and HP of milling machine and by level of milling fee.

Table 1 .
Number of rice millers interviewed by region and by year of establishment and total number of rice millers in Uganda in 2007, 2009 and 2012 by region.
Kijima et al. (2012)) surveys.2)EstimatedusingdatafromricemillsurveysbyCandiaetal.(2008),Kijimaet al. (2012)and ours.For estimation details, seeKikuchi et al. (2013).3)For09/07and12/09,compoundgrowth rate per year.4)Figures in parenthesis are the sampling ratio of rice millers interviewed.ordertoestimatethetotalnumber of rice millers in 2012, we conducted a phone survey to the District Agricultural Officers in major rice producing districts between April to August 2012.Combining data from rice mill surveys byCandia et al. (2008)andKijima et al. (2012), we estimated the total number of rice mills in 2007, 2009 and 2012 by district, which was aggregated to four regions; North, East, Central and West.3

Table 2 .
Number of milling machines by type, country made, power source, horse power and region, 2012 1) Because of missing data, some totals for #1, #2 and #3 do not tally with the total machine numbers shown at the top raw.2) According to respondents' answer, not necessarily the real origin of manufacturers.3) Figures in parenthesis are the average milling capacity (100 kg /hour) in terms of milled rice.

Table 3 .
Quantity of paddy milled, acquisition of paddy and rice millers who also operate as rice traders, 2012.

Table 4 .
Regression equations explaining milling capacity and maintenance expenditure.

Table 6 .
Data for estimating costs and returns in rice mill operation 1 .

Table 7 .
Costs, revenues, owner's surplus and rate of return for average small rice millers, 2011-2012 1 .