Demand Structure and the Consumption of Garri in Owerri North Local Government Area of Imo State, Nigeria

The study analysed empirically the demand structure and the consumption of garri in Owerri North L.G.A. of Imo state, Nigeria and sought to determine the factors that influence the monthly budget share for garri at low, medium and high income levels, estimate own price, cross price and income elasticit ies of demand for garri and as well to determine the factors that influence preference for a part icular garri brand. In order to achieve this mandate, Simple random and judgmental sampling techniques were used to select communit ies, their markets and 80 respondents. A set of pretested and structured questionnaire was used in data collection. Multiple regression model, demand elasticity indicators and logit regression model were the tools of analysis adopted for th is study. The result of the mult iple regression analysis with semi log, as lead equation showed that the critical and significant determinants of monthly budget share for garri among low income consumers were household size, price of garri and monthly incomes at given levels. The combined effect of all the variab les explained 77.8% of the variation in the monthly budget share for garri at 1.0% risk level. The result of the mult iple regression analysis with exponential functional form as the lead equation showed the critical and significant determinants of monthly budget share for garri among the middle income consumers were number of years spent in school, household size and monthly incomes. All the variables exp lained 81.8% of the variation in the monthly budget share for garri among the middle income consumers. The result also showed that the own price elasticity of demand for garri was 0.03 while cross elasticity of demand fo r garri and fufu, samovita, yam floor and cassava floor were 2.3,1.8, 1.4 and 1.4 respectively. The income elasticity of demand for garri was 0.3. The logistic regression estimat ion showed that the determinants for preference of garri include price, colour perception, age and education which were statistically significant at 1.0% risk level. It was recommended that there is the need for all consumers in different income strata (low, medium and high) to patronize garri of all colours in order to act as impetus to farmers to produce more and increase their income.


Introduction
Cassava (Man ihot spp ) is an imp o rtant crop widely cult ivated in Sub -Saharan Africa. A lthough, the crop is grown virtually in all parts of the sub-continent, production is specific in the humid t ropics [1]. Cassava plays a major role in effo rts to alleviate Nigeria's food crisis because of its efficien t p ro du ctio n o f fo od energ y , all y ear rou nd availab ility , to lerance to ext reme stress cond it ions and suitability to various farming and food systems [2]. Cassava is capable of provid ing food energy much mo re than maize, so rghu m and rice [3]. Garri, a p rod uct fro m cassava, constitutes more than 70% of total cassava consumption in Nigeria [4]. Thus, garri is one of the staple food items consumed in Nigeria. It therefore imp lies that garri is eaten by every family irrespective of their socio-economic statusand religious background. Thus, garri commands much interest and hence gets a considerable preference on the household daily budget. This means that there will be across the board high demand for the product.
Ev idence had shown that food production in Nigeria has not kept pace with the increasing population. The annual growth rate of the food sub-sector is 2.0 percent wh ile the annual population growth rate is 3.3 percent [5]. This presupposes that a wide gap existed between food supply and the demand for food by the nation's populace. The seasonality of agricultural products causes price instability. Ev idence has shown that periods of surpluses are usually supplanted by periods of scarcity. Th is vagary in the price variable cause unpredictable fluctuations in the demand and supply factors. In addition, there are other factors that Government Area of Imo State, Nigeria influence the purchase decision of consumers such as organoleptic features as colour, taste, aroma, texture, physio-chemical and hydrogen cyanide (HCN) residues.
Ezedin ma and Oti, stated that the traditional methods of cassava processing into garri are often contaminated with undesirable extraneous matters that make them unhygienic thereby affecting demand and consumption [6]. There is therefore the dire need to know the demand structure for garri at a micro setting such as Owerri North Local Govern ment area of Imo State, so as to build a bridge between, processing, availability, affordability and consumption. The significance o f this study will to a large extent contribute to good farm policy fo rmulat ion and market strategies that will enhance the growth of the food sector. Th is research is anchored on the following specific objectives; i. to determine the factors that influence the monthly budget share for garri at low, med iu m and high inco me levels in the study area.
ii. to estimate own price, cross price and income elasticities of demand for garri in the study area.
iii. to determine factors that influence preference for a particular garri brand in the study area.
The following hypothesis were tested: Ho 1 Monthly budget share for garri among lo w, mediu m and high income groups is positively related to literacy level, age, household size, price and monthly inco me.
Ho 2 Own price, cross price and income elsticit ies of demand for garri is zero.
Ho 3 Preference to a part icular garri type is positively influenced by price, colour, taste, texture, age of consumer and educational level of consumers.

Research Methodology
The study was carried out in Owerri North Local Govern ment Area of Imo State, Nigeria. The choice was purposive. This was informed by the predominance of civ il servants and existence of a large daily market in the area. The L.G.A was created in 1996 and had a population of 289,721 [5]. It lies between latitudes 4 o 5' and 6 o 5' North of Equator and Longitudes 6 o 30' and 8 o 9' east of green wich merid ian. Owerri North LGA is bounded in the North by Mbaitoli and Ikeduru LGAs and west by Owerri West and Owerri Municipal LGAs of Imo State, Nigeria [7].
Random sampling technique was used to select eight markets fro m eight co mmun ities. The eight commun ities and their markets selected were Emekuku, Awaka, Orji, Naze, Uratta Egbu and Obube. Judg mental sampling technique was used to select garri buyers fro m various daily markets in the area. Data was obtained with the use of structured and pretested questionnaire ad min istered on the respondents. Objective one was achieved using multip le regression analysis while objective two was analy zed with demand elasticity indicators. Objective three was realized with Logit regression model.
The implicit model o f the ordinary least square (OLS) mu ltip le regression is stated thus; Q = F(X 1 , X 2 , X 3 , X 4 , X 5 , ei) (1) Where Q = Monthly Budget share of garri ( N) X 1 = Years spent in school X 2 = Age of the consumer (years) X 3 = Household size X 4 = price of garri (N) X 5 = Income per month (N) The demand function was regressed using four d ifferent functional forms namely linear, exponential, double log and semi-log and the one with the best fit in terms of a priori expectation, statistical and econometric criteria was used as the lead equation.
The own price elasticity is stated thus: (2) Where: ep = price elasticity =slope of the linear demand curve Q = quantity demanded P = price of co mmodity. The cross price elasticity is stated thus: Where: Exy = cross elasticity of commod ity X with regard to commodity Y Qx = Quantity demanded of co mmod ity X Py = price of co mmodity Y The income elasticity is specified as Where Q = quantity of the goods demanded Y = Consumers inco me The Logit model is generally imp licitly specified thus Zi = bo + biXi 1 i + b 2 X 2 i + -----+ bp Xij (5) Where Zi = unobserved continuous variable fo r the ith case. Xiј = the јth coefficient (independent variables) P = number of pred ictors. This model is exp licit ly exp lained as: Zi = bo + b 1 X 1 + b 2 X 2 + b 3 X 3 + b 4 X 4 + b 5 X 5 + b 6 X 6 Zi = preference to the garri type X 1 = price of garri (N) X 2 = Colour of garri (Du mmy: wh ite I; Yellow = 0) X 3 = Taste (Du mmy: sweet taste = I; sour = 0) X 4 = Textu re (Du mmy : fine = J; Gritty = 0) X 5 = Age (years) X 6 = Nu mber of years spent in school.

Result and Discussion
The factors that influenced the monthly budget share of garri by the different inco me groups are shown in Tables I  and 2. The income groups under study were low, mediu m and high income groups. The bench mark for the stratification was according to Okorji and Ezeh that categorized monthly inco me levels of N50, 000.00 as low income, N50, 001 -N100, 000.00 as mediu m income and above N100, 000.00 was high inco me group [8; 9]. In the study area, only the low and medium inco me groups were identified and presented based on the above bench mark.
The result in Table1 shows the factors that influenced the budget share for garri among the low income households in Owerri North LGA of Imo state Nigeria. It shows that all the estimated functions were significant. ( P < 0.01). This implies that the functional forms were adequate in exp laining the variations in the dependent variables. However, among the four functional forms (Linear Exponential, Double and Se mi-Log) used for estimating the monthly budget share of garri among the low inco me consumers in the study area, the semi-log functional form was chosen as the lead equation based on econometric and statistical reasons such as the number of regression coefficients that are significant, the value of R 2 (0.778) and the significant level of F-ratio (P < 0.01).
Specifically, household size is d irectly related to the monthly budget share of the low inco me consumers. The coefficient is 2.526, wh ile the standard error is 0.952 and the variable is statistically significant at 1.0% probability level. Expectedly, consumers with large household size are likely to spend greater proportion of the monthly budget on garrithe basic staple and other basic household requirement. Although, Christianity restricts the number o f wives in the household, many low income household sizes are relat ively large. This situation has posed serious problems in recent times, due to the present economic crises and is responsible for the high rate of malnutrit ion, illiteracy and unemploy ment, especially in the rural economy [8; 9].
The price of garri made marginal contribution to the equation. The coefficient (0.976) is positive and the standard error is 0.149. The variable is statistically significant at 1.0 percent risk level. The sign of the variable is in conformity with a priori expectation in that a higher price would cause the consumer to allocate a larger proportion of the monthly budget to garri purchase. This is because higher prices cause the value of real incomes to fall, thus causing a large share of the budget to be devoted to the purchase of the commodity. But when prices fall, the value of real inco me rises, greater quantity of the good in question can be bought with reduced amount of the budget [10].
The coefficient of the inco me per month of the low income consumers (8.612) made positive contributions to the equation and the standard error is 3.768. The variab le is statistically significant at 5.0% alpha -level. The sign is in accordance with a priori reasoning. Expectedly, the positive sign imp lies that the monthly budget share of the low inco me consumers to garri increases as the monthly income increases.    The result of the mult iple regression analysis on the factors that influence the monthly budget share for garri of med iu m income consumers in Owerri North LGA o f Imo state, Nigeria is shown in Table 2. A mong the four functional forms (Linear, Exponential, Double and Semi-Log) used for estimating the monthly budget share for garri among the middle inco me consumers, the exponential functional form was chosen as the lead equation based on econometric and statistical reasons, such as the number of regression coefficients that are significant, the value of R 2 (0.818) and the significant level of the F-ratio (P < 0.01).
The coefficient of the number of years spent in school (-0.55) is marginal and negative while the standard error is 0.037. The variable is statistically significant at 90.0% confidence level. The sign of the coefficient is in conformity with a priori expectation that the monthly budget share for garri o f mediu m inco me consumers would reduce as the number of years spent in school increases. Consumers with high level of education are expected to evaluate food product by their prices and specific nutrit ive quality rather than by popularity before allocating a given share of the monthly budget to the commodity [11; 12].
The coefficient of household size is 0.252 and the standard error is 0.082. The variable is statistically significant at 99.0% confidence level. Accordingly, it is expected that middle income consumers with large household size would likely spend a greater p roportion of their monthly budget share on garri, which is the basic staple food availab le and affo rdable all year round.
The coefficient of income per month (6.993) is positive with a standard error of 1.029. The variable is statistically significant at 1.0% probability level. The sign of the variable is in consonance with a priori expectation. It imp lies that as the monthly incomes of the middle income consumers increases, the budget share allocated to garri purchase would also increase.
The values of the o wn price, cross price and income elasticities of demand for garri is shown in Table 3. The result shows that the own price elasticity of garri was 0.03. This shows that a percentage change in the price of garri would result in a less than proportionate change in the quantity demanded of it, hence being perfect ly inelastic. This result compared favourably with Effiong [13] that obtained -0.83 as own price elasticity in pork demand in Cross River State of Nigeria. Th is reinforces the readily availab ility of garri with very minimal variation in the price level.
The values of cross price elasticit ies of demand for garri with fufu, samovita, yam flour and cassava flour were 2.8, 1.8, 1.4 and 1.4 respectively. This implies that a one percent increase in price of garri, will increase the consumption of fufu by 2.3, samovita by 1.8, yam flour by 1.4 and cassava flour by 1.4 in the study area. Th is indicates a strong degree of co mpetitiveness and substitutability of these goods with garri.
The income elasticity of demand for garri is shown in Table 3. The Table shows that the income elasticity of demand fo r garri was 0.3. This imp lies that as consumers' income increase, a smaller or declining portion of their income will be spent on the product. This result is consistent with Effiong result in the consumption of pork [13]. Engel's law in terms of elasticity agrees that food is income elasticity. However, a change in consumer's money inco me will have a significant influence on consumption of the product either increasing or decreasing consumption.
The results of the Logistic regression on the factors that influence the preference of garri by consumers in Owerri North LGA of Imo state, Nigeria is shown in table 4. The logistic regression estimation for the variables showed that the effect coefficient Exp (B) for price, colour perception, age and education were statistically significant at 1.0% risk level as shown by the Wald test.
In terms of price, the result revealed that higher prices induced mediu m income consumers to have the preference and the willingness to consume wh ite garri. Similarly, the perception of colour had a positive relat ionship with preference. This shows that there exists dominant preference for white garri in the study area, contrary to the widely held scientific opinion that yellow garri contains more nutrients especially vitamin A supplied by the colour additive than the white garri. This result is inconsistent with the Umberger et al [14] assertion that quality and health peculiarit ies are important factors for preference and willingness to consume food.
With odds ratio of mo re than one, both the effect coefficient of age and education had positive significant influence on preference for white garri. By imp lication, increasing age and education impacted positively on preference and willingness to consume white garri in the study area. Older consumers with higher level of education are expected to evaluate food products by their price and specific quality rather than by popularity [11]. The Co x-Snell and Nagel Kerke R 2 values are attempting to provide a logistic analogy like the coefficient of mu ltip le determination (R 2 ) in OLS regression. At 70.4% and 94.3% for Co x-Snell and Nagel Kerke respectively, the regression line fit the data appreciably h igh and imp ly the extent of explanation of the variat ion in the dependent variable. The result consolidates the findings of Nwachukwu et al, who had similar outcome [15].

Conclusions and Recommendations
Consumers have been recurrently the unit of analysis in all economic activ ities. Each ind ividual want to maximize utilit ies but due to budget constraint must resorts to alternatives. Alternatives in terms of whether it will satisfy their need or not must be evaluated. From the analysis of the demand structure and the consumption of garri, this study reveals that the major determinants of the monthly budget share for garri among low inco me consumers in Owerri North LGA in Imo state, Nigeria with Semi-log as lead equation were household size, price of garri and monthly inco mes while the determinants of the monthly budget share for garri among middle inco me consumers with exponential functional form as the lead equation were number of year spent in school, household size, and monthly income.
The values of own price elasticity of demand for garri was 0.03 while cross price elasticit ies between garri and fufu, samovita, yam flour and cassava flour were 2.3, 1.8, 1.4 and 1.4 respectively. The inco me elasticity of demand for garri was 0.3.
The logistic regression estimation showed that price, colour perception, age and education were the critical determinants of preference for white garri and were statistically significant at 1.0% risk level.
The study recommends a policy such as sustained increase in the salary of the working class consumers by the government. Higher inco mes would increase the propensity for increased purchases hence increased income to the farmers.
All consumers in d ifferent inco me strata (low, middle and high) are encouraged to consume garri of any colour as this would be added impetus required by farmers to produce more and increase their incomes in so doing.