Appendix B: Methodology used for the design and analysis of household surveys and data.

Vulnerability: A Complex Term

There is no one single definition for the term "vulnerability" and no one single way of measuring it. Different disciplines define it and measure it differently, but the one common trend among all of them is the idea that the concept is related to levels and types of risks to which people/communities are exposed. Table B.1 summarizes some of the most commonly used definitions.

The differences between the approaches can be reduced to the tendency of each discipline "to focus on different components of risk, household responses to risk and welfare outcomes". All approaches have their strengths and weaknesses: some are considered strong in their conceptual framework but weak in their empirical approach (i.e., how it is measured) and vice versa. The definition used in this study is eclectic: it borrows from all of these disciplines.

Selection of Study Areas--Vulnerability Mapping

Vulnerability indices are commonly used in the field as a way to measure vulnerability by different researchers and institutions. Two such indices are the "Food Insecurity and Vulnerability Information and Mapping System" (FIVIMS), developed by the Food and Agriculture Organization of the United Nations (FAO); and the "Vulnerability Analysis and Mapping (VAM)", produced by the World Food Program in cooperation with FIVIMS. (58)

In this study, a vulnerability index was also developed to guide district selection. Case study sites were identified based on a vulnerability analysis using a "Principles, Criteria and Indicators" (PC&I) framework, together with a Geographic Information System (GIS). Rather than assigning weights and scores on an ad hoc basis, Principal Component Analysis (PCA)59 was employed to provide a statistical basis for determining the effect of each variable on the target variable, i.e. agricultural vulnerability. (60)

The drought- and flood-prone areas were demarcated and then overlaid with other maps containing information on other biophysical, social, and economic parameters. The basin was used as the geographical unit in the development of the maps. By superimposing maps with the different parameters and showing their fluctuation from one year to another over a reasonable period of time, a map depicting different degrees of variation is produced which serves as the basis for selecting specific sub-areas for analysis.

In this study, the secondary data on biophysical, social, and economic indicators such as gross cropped area, cropping patterns, groundwater availability, and an Infrastructure Development Index (LDI), among others, was compiled over different years spanning a 10-year period, for comparison purposes. The data was collected from various sources including the Survey of India (SOI), the Census of India (COI), the Central Ground Water Control Board (CGWB), the Central Water Commission (CWC), the National Bureau of Soil Survey and Land-Use Planning (NBSS & LUP), the National Atlas & Thematic Mapping Organization (NATMO), the Center for Monitoring Indian Economy (CMIE), the Indian Agricultural Statistics, Volumes I & II, the Agricultural Census, and the Maharashtra and National Information Center (NIC).

Using PCA, a vulnerability index was created which allocates degrees of vulnerability to districts: low, moderate, high, very high, and extremely high. Districts were classified according to the index and maps were then developed for the states of Andhra Pradesh, Orissa, and Maharashtra. An overlay of different profiles for these states thus forms the basis for the selection of the districts in each state, except for Orissa where official data was not available to allow for a comparison of vulnerability over time. Consequently, the district selection in Orissa was guided by a combination of (a) analysis of secondary data and (b) the extent of the geographical area which is considered to be liable to floods. Based on this analysis, some districts were deemed to face greater threats than others due to a combination of high biophysical and social vulnerability and limited infrastructure development.

The final selection of districts in the selected river basins was made to purposely capture a range of vulnerability patterns given their different socio-economic, technological, and biophysical conditions.

Field Surveys

Despite the fact that more than two districts in each of the three states were selected for climate projection (five in Maharashtra and four Andhra Pradesh), further prioritization of districts was necessary in conducting field surveys due to limits in time and budget. Thus, field surveys were carried out in two districts in each state. In the end, the districts of Anantapur and Chittoor in Andhra Pradesh, Jagatsinghpur and Puri in Orissa, and Ahmednagar and Nashik in Maharashtra were chosen for the study. The objectives of the surveys conducted were the following:

* to assess the coping capacities and vulnerabilities of communities in dealing effectively with droughts and floods; and

* to determine the factors that influence the effective implementation of coping measures at field level.

Institutional surveys were carried out to collect information on the central and state government plans and programs being implemented in the state and to ascertain their efficacy in enhancing the capacities of communities in dealing effectively with climate variability and conditions of extreme weather, including drought and floods. The field surveys sought to collect information on the communities' perceptions on (a) the intensity of droughts/floods, (b) the crops grown in the region, (c) the change in cropping patterns, irrigation, livelihood options and migration, (d) infrastructure, (e) the availability of financial services and schemes, and (f) the importance of insurance. Through these surveys, an attempt is made to undertake a critical review of policy and community-oriented interventions that enhance the capacities of communities to cope during extreme climate situations. In all, 1,640 households were surveyed: 570 households in Andhra Pradesh, 650 households in Orissa, and 420 households in Maharashtra.

Development of Tools for Institutional and Field Surveys

Questionnaires designed for implementation in drought and flood circumstances as well as other Participatory Rural Appraisal (PRA) tools were used. Secondary data including sketch maps, transect walk, collation of time-line information and trend-lines, seasonal cropping calendar mapping, institutional mapping, problem tree analysis, and problem and opportunity ranking was collected. In addition, group discussions, interviews, focus group discussions, and institutional surveys were carried out.

The questionnaires were pre-tested in pilot surveys in Rajasthan (a drought-prone area). It provided insights about the available quantitative information and its usefulness for the purpose of the survey, and it was improved and modified accordingly.

The lack of proper recorded information at the village level posed a major constraint to quantitative/statistical analysis.

Selection of Villages in Identified Districts Based on Analyses of Secondary Data

The selection of villages in each district was based on the screening of village-level secondary data collected from the census office. This data was collected for parameters including village land area, land use, cultivated and irrigated land, and availability of infrastructure including education, bank/credit, society, communication, power facility, and services like health care. The data was used for the preliminary selection of villages within each district. These were later confirmed by discussions with officials in government departments at the district level as well as other localized non-governmental organizations and communities at the village level.

Village Classification Based on Irrigation

All villages lying within a district were classified into one of three levels based on their irrigated area as a percentage of their total agricultural area: low (0-33%), moderate (33-66%), or high (66-100%).

Village Classification Based on Infrastructure Development

An infrastructure index was developed by considering the existence or level of certain facilities and services at the village level including the availability of drinking water, education facilities, medical facilities, electricity, banks, agricultural society, and communication linkages.

The villages were assigned to one of four categories according to their irrigation- and infrastructure-based classification. The purpose of this categorization was to select villages that were representative of different contexts which may further the understanding of the factors underlining the different levels of vulnerabilities. These broad criteria on irrigation and infrastructure are used to classify the villages in a matrix as the one shown here below.

Sampling

The sample size, n, for each target population is computed using the formula by Murthy, (1977):

n = N * n/N + n -1

Here, n = [c.sup.2]/[e.sup.2] where c is the population coefficient of variation and e is the allowed percentage of error. ./Vis the target population size (480).

To obtain a representative sample, proportionate sampling based on landholdings were conducted. Records indicating household land category were collected from the Tehsildar (61). The survey was conducted based on the various landholding categories: >4 acres (large farmers); 1-4 acres (medium farmers); < 1 acre (small farmers) (62); and landless. After conducting the survey, all the data was coded and entered, and were used for quantitative analysis. A reference manual was also developed to facilitate viewing and referencing.

Measure of Income Volatility

The Coefficient of Variation (CV) is used to understand the extent to which household incomes are volatile to the impact of drought/floods events (63). The CV is simply a measure of the deviation of 'impact year' income from 'normal year' income.

CV is defined as the ratio of the Standard Deviation to the Mean ([mu]) and is defined by the following formula:

[c.sub.v] = [sigma]/[mu]

In this formula, [sigma] is standard deviation and [mu] is average income (64). It is often represented as a percentage by multiplying the above by 100.

An advantage of the CV is that it is free from the units of the variables, and it thus permits comparisons with respect to their variability. The CV is commonly used since it is a quantity without physical units. Although the CV indicates the magnitude of variations, it fails to capture the directional shifts in income. As a substantial majority of the surveyed households experienced drops in income in an impact year, few 'outlier' households that showed an increase in income during an impact year were segregated out.

Table B.1: Definitions of Vulnerability

Discipline/
Literature     Definition of Vulnerability

Economics      It is an outcome of a process of
(57)           household responses to risks,
               given a set of underlying
               conditions. Often times, the
               outcome is poverty.

Sustainable    It is the probability that
Livelihoods    "livelihood stress" will occur--
               with more stress or a higher
               probability implying increased
               vulnerability. Also, "the balance
               between the sensitivity and
               resilience of a livelihood
               system."

Food           It is the risk of irreversible
Security       physical or mental impairment
               due to insufficient intake of
               macro or micronutrients.

Disaster       It is the characteristics of a
Management     person or group in terms of their
               capacity to anticipate, cope with,
               and recover from the impact of a
               natural disaster. It is an
               underlying condition separate
               from that of the risky events that
               may trigger the outcome. It
               refers to risks as "hazards".

Discipline/    What is Measured How
Literature     it is Measured

Economics      The fall of income beyond
(57)           the poverty line or
               changes in consumption
               are measured.

Sustainable    The loss of livelihood,
Livelihoods    continued vulnerability to
               subsequent shocks and
               vulnerability changes over
               time are the subjects of
               interest. The assessments
               are specific to population
               or society. It uses a case
               study approach.

Food           Vulnerability mapping
Security       and indexes. A number of
               analytical techniques are
               used to examine the
               degree of correspondence
               between the concept of
               food security and the
               indicators chosen to
               measure it.

Disaster       Vulnerability =
Management     Hazard--Coping
               Household characteristics
               are key determinants in
               that they affect either side
               of the equation.
               The use of vulnerability
               mapping is also
               widespread.

Discipline/
Literature     Criticism

Economics      There is an underlying
(57)           presumption that all
               losses can be measured in
               monetary terms.

Sustainable    It tends to use terms and
Livelihoods    concepts that are unclear
               or not widely accepted. It
               is not clear how changes
               in vulnerability would be
               evaluated over time when
               some indicators show a
               positive change while
               others a negative one.

Food           It usually lacks a
Security       benchmark to which
               indicators can be
               compared. It recognizes
               that vulnerability is made
               up of different
               components, but it ignores
               the specific process by
               which the components
               interact to determine
               overall vulnerability.

Disaster       There is a lack of
Management     precision in the language
               used, which leads to
               confusion. At times, it
               fails to be specific about
               what constitutes loss or
               damage, or whether it
               matters who endures
               these.

Figure B.1 Example of the Village Classification in the
Infrastructure-Irrigation Matrix

                  High Irrigation       Low Irrigation

High              For instance:         For instance:
Infrastructure    Korhate in            Manesaudram in
                  Maharashtra           Andhra Pradesh

Low               For instance:         For instance:
Infrastructure    Neramatla in Andhra   Brahmanapalle in
                  Pradesh               Andhra Pradesh