• Keine Ergebnisse gefunden

We conclude this introductory chapter with a brief discussion on the main sources of information used in analysis of the Malawi subsidy programme.

We draw on four main sources of information:

• implementation records on the subsidy programme;

• household and input supplier surveys conducted in 2006/7, 2008/9, and 2010/11 as part of the evaluation of the programme;

• offi cial statistics;

• other studies on the subsidy programme.

We discuss each of these in turn.

1.3.1. Implementation records on the Malawi subsidy programme

Since 2006/7 the logistics of subsidized fertilizer distribution and payments to fertilizer suppliers, fertilizer transporters, and seed suppliers have been managed by the Logistics Unit, working in close cooperation with the Ministry of Agriculture and Food Security (MoAFS), donors, the two par-astatals involved in subsidized fertilizer and seed distribution (Agricultural Marketing and Development Corporation and Smallholder Farmer Fertilizer Revolving Fund of Malawi, ADMARC and SFFRFM), and contracted trans-porters and seed and fertilizer suppliers. In Chapter 5 extensive use is made of information from the Logistics Unit’s weekly and annual reports, sup-ported by minutes of weekly task force meetings and information supplied directly by the MoAFS.

1.3.2. Programme evaluation studies

Much of the information and analysis on implementation is also contained in various reports of FISP evaluations led by the authors (for example, SOAS et al., 2008; Dorward et al., 2010; Dorward and Chirwa, 2011). Since 2006/7 the authors have led annual evaluations of the subsidy programme, with more intensive and ‘light touch’ evaluations in alternate years. More inten-sive evaluations of the 2006/7, 2008/9, and 2010/11 programmes involved household surveys with focus group discussions and a community survey and in 2006/7 and 2008/9 an input supplier survey. ‘Light touch’ evaluations of the 2007/8, 2009/10, and 2011/12 programme years have drawn mainly on implementation records as outlined above, together with information from other studies and offi cial statistics, and analysis of data from more intensive evaluations.

The 2006/7 survey used a sub-sample of households sampled in the National Statistical Offi ce (NSO) 2004/5 second Integrated Household Survey (IHS2) in order to provide panel data for analysis of programme impacts on benefi ciar-ies. A total of 3298 households were sampled across all districts in Malawi.

After data cleaning this gave 2431 balanced matched panel households also sampled in the IHS2. Agro-economic livelihood zones defi ned by the Malawi Vulnerability Assessment Committee (MVAC) were used to stratify the sam-ple (Malawi National Vulnerability Assessment Committee, 2005). Urban, peri-urban, and protected areas (national parks and reserves) were omitted from the sample. Data collection and entry were conducted by the National Statistical Offi ce. The survey provided very valuable information on house-hold access to subsidized and unsubsidized inputs and on many aspects of programme implementation. Unfortunately it was less successful as regards plot areas and production reported by farmers: these were not found to give

reliable and consistent results, and this prevented estimation of production impacts of the programme. The 2006/7 input supplier survey involved focus group discussions, key informant interviews, and a survey of 271 retail out-lets in 6 districts. This was supplemented by information from fertilizer and seed importers and sellers. The fi ndings were reported in School of Oriental and African Studies et al. (2008).

The household survey in 2008/9 was conducted by the evaluation team with a sample of 1982 households across 14 districts and represented almost all livelihood zones. The sample was a sub-set of the 2006/7 sample and there-fore provided a panel data set across three surveys going back to the IHS2.

The input supply retailer survey sampled 230 retailers in 6 districts. Both surveys were again supplemented by focus group discussions, key inform-ant interviews, and a community survey, but detailed fertilizer import infor-mation was not available. Findings were presented in a portfolio of reports focusing on different aspects of the programme (for example, Dorward et al., 2010a, b; Kelly et al., 2010). The survey again provided valuable information on programme implementation and outputs. However, the introduction of innovative approaches to production and yield measurement (such as yield sub-plots with enumerator and farmer harvests), plot areas, and production data did not give reliable and consistent results. This not only precluded estimation of production impacts of the programme, it also raised questions about the reliability and consistency of area, production, and yield estimates from other studies which rely mainly on farmer estimates and recall of pro-duction (see Dorward and Chirwa, 2010b).

The 2010/11 study did not include an input supplier survey, and the sam-ple size of the household survey was reduced further to 760 households across 8 districts in the 3 regions. The sample represented 8 major maize growing livelihood zones covering 77% of all rural households and was again a sub-set of the previous survey (this time the 2008/9 survey). The 2010/11 survey replaced attrition households with younger and newly formed households.

The IHS2 and three FISP evaluation surveys generated a matched panel of 461 households. As for 2008/9, fi ndings were reported in a portfolio of topic-spe-cifi c reports (for example, Chirwa et al., 2011d; Dorward and Chirwa, 2011a;

Mvula et al., 2011).

A number of specifi c studies were conducted and reported within the programme evaluation—for example, on programme impacts, benefi t–cost analysis, targeting, and graduation. These are explained and cited where appropriate. It is, however, necessary to provide a little more information here on the development and use of the partial equilibrium Informal Rural Economy or IRE model to explore some of the economy-wide impacts of the programme. This model is fully described in Dorward and Chirwa (2012b). It is based on detailed programming models of different farm household types

in the two most populous livelihood zones in Malawi. These models allow for seasonal constraints affecting farm household activities and the direct impacts of the subsidy are investigated by simulating the livelihood effects of specifi c households’ access to subsidized inputs. These effects are then aggre-gated in order to estimate impacts on supply and demand of seasonal labour and maize. Wage rates and maize prices are then adjusted iteratively to fi nd new equilibrium wages and prices and to derive estimates of economy-wide impacts on both subsidy recipients and non-recipients. The nature of the data available and of the models means that results should be interpreted as indic-ative of possible effects rather than predictive of actual effects. Nevertheless, when taken together with other information they provide useful insights into possible economy-wide impacts.

A full set of evaluation reports from 2006/7 to 2011/12 can be found at

<http://www.wadonda.com/ > . These provide further documentation of analytical methods and references are provided whenever their fi ndings are drawn upon.

1.3.3. Offi cial statistics

Malawi has an extensive set of agricultural and other national statistics.

The Ministry of Agriculture publishes valuable monthly information on market prices for major crops, with data collected on a weekly basis from a large number of markets around the country. The Ministry also publishes annual estimates of crop areas, production, and yields, and reports annual estimates of the number of farm families. The annual production estimates, with large increases in estimated maize production following the introduc-tion of the subsidy programme in 2005/6, have been widely cited as evidence of the impact of the subsidy programme. However, there are inconsistencies between the large estimated production increases from 2005/6 and the very high domestic maize prices experienced in some years, notably in 2008/9.

These inconsistencies are discussed in Chapter 7. There are also inconsisten-cies between national maize areas, production, and yield estimates from the Ministry of Agriculture, and from different reports by the National Statistical Offi ce (National Statistical Offi ce, 2005a, 2010a). These discrepancies are dis-cussed in more detail in Dorward and Chirwa (2010b) and summarized in Chapter 6. Another set of discrepancies between Ministry of Agriculture and NSO data concerns the number of farm families (reported annually by the Ministry of Agriculture) and the number of rural households enumerated in the 2008 census (National Statistical Offi ce, 2008a). This discrepancy and the diffi culties it raises are discussed in Chapter 5.

Apparent discrepancies also arise between maize prices reported by the Ministry of Agriculture and the consumer price index reported by the NSO.

No detailed analysis of this has been published, to our knowledge, but the high maize prices observed in 2008/9 and in 2011 do not appear to be con-sistent with consumer price index fi gures for the same period, given the high weighting of food and particularly maize in the consumer price index. We also note that the NSO itself refers to revised ‘CPI data’ with ‘overall infl ation between the IHS2 and IHS3 periods of 128.9 per cent’ (National Statistical Offi ce, 2012: p. 207), when offi cial CPI estimates for the same period suggest a considerably lower rate of infl ation. This also raises wider questions about the defl ator used in recent years’ GDP estimates, and hence about these GDP estimates themselves.

Integrated Household Surveys (IHS) conducted by the NSO in 2004/5 (IHS2) and 2010/11 (IHS3) provide national estimates on a wide range of variables.

We refer to these in Chapters 4 and 7. However, we also note some apparent discrepancies between and within some of the results presented, and—with the publication of the fi rst report on the 2010/11 survey (National Statistical Offi ce, 2012) as the manuscript for this book was being fi nalized—it has not been possible to resolve these.

1.3.4. Other studies on the subsidy programme

A number of other studies have been made of different aspects of the subsidy programme. Due to their varied nature and focus we do not discuss these here but refer to them at relevant points in the following chapters. In broad terms the main focus of most other work has been to use survey data to compare observations on recipients and non-recipients in order to examine targeting of and direct outcomes and impacts of subsidy receipt—in terms of differences in wealth, gender, and other household characteristics affecting access to and use of subsidized and unsubsidized inputs, and subsequent differences in changes in wealth and other household characteristics. There has been much less examination of indirect and economy-wide impacts, of impacts on mar-ket development, of benefi ts relative to costs, and of the important question of graduation, though we discuss notable exceptions where appropriate.

Part I

Background

In this part of the book we provide essential theoretical and empirical context for the rest of the book. The three chapters

• set out and extend more conventional theories on agricultural input subsidies’ strengths and shortcomings;

• review limited information available on twenty-fi rst century agricultural input subsidy programmes in sub-Saharan Africa; and

• describe Malawi’s specifi c political, economic, and agricultural features.

These chapters structure and underpin the description, analysis, and evalu-ation of the Malawi subsidy programme, and the wider applicevalu-ation of that evaluation, in the remainder of the book.

2

Agricultural input subsidies: changing theory

Im Dokument Agricultural Input Subsidies (Seite 24-31)