6.2 Supply Response

Both of the villages included in the survey seem to have benefited from the price and marketing reforms undertaken as a part of Uganda’s structural adjustment programme: the nominal price of coffee increased from Shs 120/kg in 1990 to Shs 700/kg in 1997, while the nominal price of maize increased from Shs 30/kg to Shs 450/kg over the same period (Government of Uganda, 1996/97). The price data collected in the survey, while not definitive, suggest that farmgate prices in the selected villages were only slightly lower than these national-level data would suggest.

It is very difficult to estimate a national-level supply response to NTAE promotion policies for several reasons. First, there are no reliable recent data on total agricultural output. In addition, while some unofficial cross-border trade has presumably been switched to official trade, no estimate exists on past or current unrecorded exports. Finally, there are no reliable estimates of the proportion of crops marketed domestically. It was not possible to ascertain what proportion of staples is produced for own household consumption or is sold locally.

Unfortunately, the prices for most crops grown in Gonve and Kinyatta changed little over the two survey years (1996 and 1997), and thus very little in the way of a quantitative estimate of a local-level supply response can be derived from the survey data. However, the survey and the PRA exercise did yield a significant amount of information on the conditions necessary for a positive supply response to be obtained. This is discussed below.

The surveys in Gonve and Kitanyatta showed some evidence of limited increased production over the survey years. Forty-nine households in Gonve, or 25 per cent of those surveyed, indicated that they had increased labour demands due to increased production over the last season, and 26 households in Kitanyatta, or 13 per cent, reported the same. The data on acreage were not considered reliable enough to draw firm conclusions about changes in cropping area over the two seasons.

The survey gave some indications of factors entering into smallholders’ production decisions (tables 11 and 12). Pricing was clearly important for cropping decisions, although it was not the only factor. Confidence in markets was also important, because without this smallholders have little faith in pricing projections. In addition, concern for food security was evident; this also limits supply responses to price changes.

Table 11
Percentage of households increasing or decreasing crops: Gonve (survey data)

Crop	% of hh increasing	Reasons	% of hh decreasing	Reasons
Maize	9	previous famine	8	pests
		good price		lack labour
		ready market		other
Beans	13	previous famine	9	lack labour
		good price		other
		ready market
Vanilla	15	good price	9	spoils other crops
		ready market		lack labour
Cassava	22	ready market	12	pests
		previous famine		other
		good price
Coffee	20	good price	4	other
Others*	12	ready market	17	bad season
		easy to grow		other

* Primarily groundnuts and vegetables.

Table 12
Percentage of households increasing or decreasing crops: Kitanyatta (survey data)

Crop	% of hh increasing	Reasons	% of hh decreasing	Reasons
Maize	16	good price	15	lack of seed
		previous famine		bad weather
		ready market		pests
Beans	4	good price	19	bad weather
		previous famine		lack of seed
		ready market		pests
Cassava	18	good price	10	bad weather
		previous famine		lack of seed
		ready market		pests
Coffee	1	no reason given	1	no reason given
Others*	10	good price	13	pests

*Primarily groundnuts and vegetables.

These data suggest that improved pricing, improved markets and increased food security would trigger production increases for most crops. However, it cannot be determined whether an increase in production in any particular crop would be at the expense of another crop - that is, whether total agricultural production would increase, or whether crops would simply be switched. There is some suggestion of a tendency to switch from beans to cassava in Kitanyatta, while in Gonve, the more prosperous village, there is more indication of a slight increase in total production. Constraints on labour and other resources would limit the opportunities for increasing total production; the extent of these constraints in the survey villages is discussed further below.

It was also evident from the PRA data that the factors necessary for production increases - good prices, ready markets and food security - were far from assured in the study villages. In addition, the labour and technological constraints that were identified as reasons for decreasing production were clearly pervasive problems (table 13).

Table 13
Problems limiting productivity identified by farmers, in order of importance (PRA data)

	Gonve	Kitanyatta
Women	1. no market for our crops	1. no money to hire labour
	2. no new seed varieties	2. no tractors for hire
	3. no agricultural officers to consult	3. most time spent looking after families
	4. old unproductive soils	4. no market for our crops
Men	1. low prices for our crops	1. lack of a viable market for our crops
	2. lack of credit facilities	2. bad feeder roads
	3. limited farm implements	3. tractors and oxen too expensive
	4. lack of extension and veterinary services in our village	4. inaccessibility to loans and credit schemes

Both men’s groups considered lack of adequate sales opportunities as the biggest problem limiting their productivity. While women in Gonve are also concerned with marketing, women in Kitanyatta show clear evidence of labour constraints. Of note is the fact that neither women’s group mentioned inaccessiblity of credit as a problem. Women’s lack of interest in credit has been noted in other contexts in which women are primarily responsible for providing staple foods: because they must provide basic needs, they must behave in a very risk-averse manner, which in large part precludes involvement in credit schemes.

Tables 14 and 15 show what women and men believe to be the causes of the problems they identified, and how they cope with these problems.

These tables suggest that the problems confronting men are to a large extent the ones that have been identified and targeted by government programmes for improving the marketing of agricultural products: high costs of transportation, poor storage facilities, little access to credit. It is striking that, in comparison to men, women tend to face problems more in the realm of the process of agricultural production itself: they emphasize poor agricultural technology, poor seeds and soils, and, above all, continual labour constraints.

The qualitative survey and PRA data thus show some potential for a positive supply response to price movements and marketing improvements. This information also makes clear the extent of the problems that smallholders face, and the paucity of resources and coping strategies they have for dealing with these problems. Finally, these data suggest the complexity of the agricultural productivity problem, and the need for policies that deal simultaneously with multiple facets of agricultural production and marketing.

Table 14
Perceived causes of problems, Gonve and Kitanyatta: Men and women (PRA data)

Problem	Perceived causes (men)	Perceived causes (women)
Poor marketing of our crops (low prices and lack of market)	· no competition among buyers · no communication with external market · lack of co-operation among farmers · buyers cheat when weighing produce · government does not care to look for markets for our crops · we have bad feeder roads · no storage facilities, so we sell immediately after harvest when prices are low	· each farmer markets individually · each farmer grows little of many crops · lack of additional labour to produce more · we have no women’s leader who is educated enough to give us new ideas · too many middlemen
Lack of implements	· petrol is very expensive so we cannot hire tractors · our produce is bought at low prices · proceeds from our crops are used for very many things · farm implements are very expensive	· large scale farmers own the tractors and do not want to help the ordinary person · men do not allow us to form groups to hire tractors · we cannot afford implements
Poor extension services	· government workers do not like villages · extension workers do not have transportation · the vets sell animal drugs at high prices · extension workers no longer have demonstration gardens in the villages	· government workers only stop at the sub-county headquarters and select a few people to train · politicians who would help us only come when they are looking for votes
Lack of access to loans and credit schemes	· there are no clear processes for farmers to access loans · the money earmarked for farmers does not get to them · our leaders swindle the money meant for loans · the interest charged on loans is more than the farmers can afford · farmers are looked down upon as illiterate · there are too many things to go through before one gets a loan
Most time is spent looking after families		· some of us are widows · lack of co-operation with husbands (all domestic work is left to women) · husbands are the decision makers, even over their wives’ money · husbands control women’s movement · it is women who care for children
Poor feeder roads	· government does not care about farmers · workers who repair roads are not paid · road maintenance equipment is used for private work · because there are no trustworthy leaders, contracts for maintenance are awarded to incompetent people
No money to hire labour for farming		· most time is spent looking after family, especially children · low prices for crops · we plant late due to lack of tractor · women are not decision makers in farming and marketing · we spend more time on the men’s gardens · we are responsible for producing food for the whole family · there are no alternative ways of making money apart from farming · we are being ruled by husbands so we cannot make a lot of money
No new seed varieties		· we get no visits from agricultural officers · we do not know where to get seeds
Old unproductive soils		· continuous cropping on same piece of land · no labour to open new land · men care for coffee plantations only · no advice on new farming practices to increase yield

Table 15
Problems and coping strategies, Gonve and Kitanyatta: Men and women (PRA data)

Problem	Coping strategies (men)	Coping strategies (women)
Poor marketing of crops	· we just sell to any buyer at very low prices · we have no choice but to accept being exploited	· we sell to whoever comes · we keep telling visitors about our problems hoping they will tell government · we sell our produce in small quantities at the local market
Lack of implements	· we use hand hoes and grow little	· we use hand hoes · we hire casual labour once in a while
Poor extension services	· we just go without them	· we use traditional methods of farming
Lack of access to credit	· we try to argue that credit facilities should be decentralized to communities without much success	· we plant only what we can manage with our own hands
Most time is spent looking after families		· we have no solution but to work hard · we report our husbands to local leaders
Poor feeder roads	· we do not travel much, we wait for buyers to find us at our homes
No money to hire labour for farming		· we only plant what we can manage · sometimes children help
No new seed varieties		· we keep replanting our own seed
Old unproductive soils		· we rest the land from time to time · we make do with our low yields · we reduce the number of meals a day in time of scarcity

Given the suggested potential for a positive supply response in the PRA data, the survey attempted to ascertain more exactly the supply response in the two villages. However, as will become clear, problems with the data - especially price data - limited the extent to which this could be done.

Prices for most of the crops produced in the two villages changed very little over the course of the survey. Aggregate national data for crop prices are available (table 16), but they are not complete. In particular, prices for most crops for the years 1995 and 1996 are not available, since surveys were not carried out for these years. This is unfortunate, as these are the key prices for this study, which looks at supply response between 1996 and 1997.

Table 16
National crop prices, 1993-1998 (Shs/kg)

Commodity	1993	1994	1995	1996	1997	1998
Maize	125	160	-	-	450	-
Beans	250	500	-	-	800	-
Vanilla	-	-	4,000	3,000	2,500	2,500
Cassava	65	100	-	-	300	-
Coffee	250	300	-	-	700	-

Data provided by the following: for vanilla, Sekalala Enterprises; all other crops, Agricultural Policy Secretariat, Ministry of Planning and Economic Development.

Because there are variations in prices paid to farmers at the farm gate, the survey was also intended to provide information about prices received by individual farmers or households. This, as with the cropped area, is a derived figure based on reported sales and volume of crops sold. The resulting price per unit data do not look particularly robust. However, the modes agree quite well with the national data for 1997, although being slightly lower.

Three types of indicators for the supply of crops were derivable from the data set: self-assessed crop increase or decrease from 1996 to 1997; an estimate of changes in the area under crops; and changes in sales of crops. Problems with the quality and consistency of these data mean that quantitative estimates of changes in production cannot be made. However, the data give an overall impression of little shift in production between years. More than 75 per cent of households reported no change in cropping, no change in acreage under production, and no change in crop sales. Among the households reporting changes in these variables, almost equal numbers reported increasing production and decreasing production of each crop.

This stasis of production is not too surprising, given the absence of price changes. In addition, the large proportion of “no change” observations in the supply response variables means that regression equations used to draw out supply relationships are likely to be of limited use, since regression is best used where there is a more even distribution of the dependent variable.

Response by type of household

Despite the data limitations noted above, an attempt was made to test for different patterns of production behaviour in different types of households (male- and female-headed, monogamous and polygamous). These tests yielded few useful results. Taken in relation to minimal average price movements between the two years, the results show, as expected, only marginal changes in planted area on average. It is difficult to identify a pattern, either by crop or by type of household. The strongest positive movements are for maize and cassava are in polygamous households, the strongest negative movement is for beans in female-headed households. For the higher value cash crops of vanilla and coffee, the strongest change in area comes in male-headed monogamous households. Interestingly, for the only crop for which the modal price increases between 1996 and 1997 - beans - the response of all types of households is to decrease area planted.

The other measure examined here is crop sales. Again, patterns are hard to identify. Overall, all types of household are selling less maize and beans, and more vanilla. For coffee and cassava there are different responses. Polygamous households seem to be changing their sales of crops the most, especially for coffee. Female-headed households appear to be selling more cassava in 1997.

Supply response for coffee and vanilla by type of household

With coffee and vanilla, sufficient price data exist to allow a more sophisticated analysis. Table 17 shows the results of a simple bivariate correlation of price changes with changes in area planted, while table 18 shows the correlation with volumes marketed. In table 17, none of the Pearson correlation coefficients are significant. The small number of female-headed households growing coffee make correlation in this case impossible.

Table 17
Correlation of price changes with changes in area planted, 1996-97

	Vanilla	Coffee
All households	0.024	-0.131
Female-headed households	-0.322	-

Table 18
Correlation of price changes with changes in crops marketed, 1996-97

	Vanilla	Coffee
All households	-0.193	0.019
Female-headed households	-0.586	-0.007

The correlation results imply first of all a generally low level of relationship between prices and supply, with the exception of female-headed households’ marketed supply of vanilla (the only significant result). In the case of vanilla, all households appear to have a negative supply response; for coffee the correlations are more ambiguous.

A more general analysis of supply response, taking into account non-price factors, was also attempted. To identify the factors determining whether households enter into the cultivation of a particular crop, probit equations were estimated. The explanatory variables were socioeconomic status, the sex of household head, the total area cultivated (as a measure of wealth), a vector of dummies for input use (fertiliser, insecticide, improved seed, and tractor hire), the use of hired labour, and the number of female and male adults in the household. These equations were estimated for both coffee and vanilla. Overall, these equations performed poorly. Both were not significant overall, and although a few of the individual variables were significant, the sex of household head was not.

Finally, for those households growing vanilla or coffee, linear equations for supply response were estimated. The equations included price changes as an explanatory variable.² In order to understand the different supply responses of male-headed and female-headed households, it is necessary to estimate the supply response equations separately for the two groups. However, there are so few cases of female-headed households growing these crops in the sample that separate estimation is impossible. Instead, dummies for sex of household head and polygamy were entered as variables in equations estimated for the sample of all households growing the crops. This is not ideal, but is the best that can be done.

² Note that these equations were estimated only for households already growing the crop, as opposed to the entire sample. This is because the behaviour of households not growing the crop, in response to various factors, is not observed, but rather their response is entered as zero. This would bias parameter estimates downwards. The standard solution to this problem is the use of the tobit model. However, this was not available, making the above approach necessary.

The results are shown in table 19, for both crops and using change in area and change in volume of crop marketed as dependent variables. The supply response analysis includes both the price of the crop and of the other main non-food cash crop. “Normal” supply response would have a positive sign for the parameter on the own-price variable, and a negative sign for the other crop. Use of hired labour and family labour, along with land area and a combination of inputs (including fertiliser and tractor use) are included. Finally, there are dummy variables for whether the household is female-headed or polygamous.

Table 19
Supply response for coffee and vanilla: Regression results

		Dependent variable
		Coffee		Vanilla
Independent variable		Change in area	Change in crops marketed	Change in area	Change in crops marketed
Constant		4.5	1638.12	1.5	350.69
Coffee price change		s	-348.8***	-0.11	-4.7
Vanilla price change		1.7*	-42.9	1.2	95.1
Hired labour (female)		0.410	-672.8	-0.03	-111.5
Hired labour (male)		-0.224	319.5	0.17	70.2
No. females		0.33*	-45.1	0.2	-14.9
No. males		-0.229	-3.2	-0.2	26.1*
Total land area		0.067	-2.4	-0.001	2.9
Inputs		-0.456	-76.3	-0.154	-27.3
Female-headed		-0.165	202.0	0.05	-2.9
Polygamous		0.403	254.9	1.08	134.5*
Adj. R2		.321	.339	-.072	-.117
F		2.135	2.54	0.84	0.77

* significant at the 10 per cent level
** significant at the 5 per cent level
*** significant at the 1 per cent level

Generally the equations for coffee perform rather better than those for vanilla, with adjusted R²s in the low 0.3 range. Only one variable is strongly related to a supply indicator, which is the coffee price for the volume of crops marketed. This is strongly negative. Female-headedness and polygamy are associated positively with supply changes, but not significantly so. Coffee production seems to involve family female labour more than vanilla does, a result not particularly borne out in the labour data below.

The results on vanilla in the correlation analysis (i.e. a negative supply response) are not borne out here. The only significant variables are for the change in crops marketed, where the number of adult men in the household and polygamy are positively related to a larger supply response.

In conclusion, the quantitative supply response analysis is hampered by the fact that there was little price movement between the two years of the study. The available data show no clear patterns distinguishing female-headed from male-headed households in their supply response behaviour, nor monogamous from polygamous households. This does not necessarily mean that no such patterns exist: the absence of price changes and non-robust production data make it impossible to generalize from these results. It is interesting, however, that there is no suggestion of a positive supply response, and several suggestions that supply response to increased prices might in fact be negative. This possibility will bear a closer analysis, especially in light of survey data from Zambia with similar findings (Wold, 1997). Given the clear interest shown by smallholders in improved prices and markets, the reason for a negative supply response is likely to lie in constraints on production, and labour appears to be the most binding constraint facing smallholders.