I. The construction of the welfare indicator:
Poverty analysis requires three main elements. First, a measure of welfare that is both measurable and acceptable, and that will allow us to rank all population. Second, an appropriate poverty line to be compared against the chosen indicator in order to classify individuals in poor and non-poor. Lastly, a set of measures that combine individual welfare indicators into an aggregated poverty figure.
Here, it explains all the steps involved in the construction of the consumption measure, the derivation of the poverty line and the poverty measures.
1. The choice of the welfare indicator
Poverty involves multiple dimensions of deprivation, such as poor health, low human capital, limited access to infrastructure, malnutrition, lack of goods and services, inability to express political views or profess religious beliefs, etc. Each of them deserves separate attention as they summarize different components of welfare, and indeed may help policy makers to focus attention on the various facets of poverty. Nonetheless, often there is a high degree of overlapping: a malnourished person is also poorly educated and without access to health care.
Research on poverty over the last years has reached some consensus on using economic measures of living standards and these are routinely employed on poverty analysis. Moreover, income-based poverty indicators are the basis to monitor the first of the Millennium Development Goals. Although they do not cover all aspects of human welfare, they do capture a central component of any assessment of living standards. The main decision is to make the choice between income and consumption as the welfare indicator. Consumption is the preferred measure because it is likely to be a more useful and accurate measure of living standards than income. This preference of consumption over income is based on both theoretical and practical issues.
The first theoretical consideration is that both consumption and income can be approximations to utility, even though they are different concepts. Consumption measures what individuals have actually acquired, while income, together with assets, measures the potential claims of the person. Second, the time period over which living standards are to be measured is important. If the interest is the long-run, as in a lifetime period, both should be the same and the choice does not matter. In the short-run though, say a year, consumption is likely to be more stable than income. Households are able to smooth out their consumption, which may reflect access to credit or savings as well as information on future streams of income. Consumption is also less affected by seasonal patterns than income, for example, in agricultural economies, income is more volatile and affected by growing and harvest seasons, hence relying on that indicator might over or underestimate significantly living standards.
On the other hand, there are practical arguments to take into account. First, consumption is generally an easier concept to grasp for the respondents rather than income, especially if the latter is from self-employment or own-business activities. For instance, workers in formal sectors of the economy will have no problem in reporting accurately their main source of income i.e. their wage or salary. But people employed in informal sectors or in agriculture will have a harder time coming up with a precise measure of their income. Often is the case that household and business transactions are intertwined. Besides, as it was mentioned before, seasonal considerations are to be included to estimate an annual income figure. Finally, we also need to consider the degree of reliability of the information. Households are less reluctant to share information on consumption than on income. They may be afraid than income information will be used for different purposes, say taxes, or they may just considered income questions as too intrusive. It is also likely that household members know more about the household consumption than the level and sources of household income.
2. The construction of the consumption measure
Creating an aggregate of consumption is also guided by theoretical and practical considerations. First, it must be as comprehensive as possible given the available information. Omitting some components assumes that they do not contribute to people's welfare or that they do no affect the rankings of individuals. Second, market and non-market transactions are to be included, which means that purchases are not the sole component of the indicator. Third, expenditure is not consumption. For perishable goods, mostly food, it is usual to assume that all purchases are consumed. But for other goods and services, such as housing or durable goods, corrections have to be made. Lastly, the consumption aggregate comprises five main components: food, non-food, housing, durable goods and energy. The specific items included in each component and the methodology used to assign a consumption value to each of these items is outlined below.
The food component can be readily constructed by simply adding up all consumption per food item, normalized to a uniform reference period, and then aggregating all food items per household. HIES records information on food consumption at the household level for 92 items, organized in 10 categories: meat and meat products, milk and dairy products, flour and flour products, vegetables, fruits, sweets, tea, coffee and beverages, spices, alcohol and tobacco, and meals eaten away from home. The information on HIES was collected through a diary left to the household for three consecutive months, enumerators went to the household at the end of each month and based on the diaries, they filled out the questionnaires. Theoretically speaking then, the food component uses factual data from a 3-month period as opposed to the typical last week or last month recall period.
A few general principles are applied in the construction of this component. First, all possible sources of consumption should be included. This means that the food component shall comprise not only expenditures on purchases in the market or on meals eaten away from home but also food that was own produced, received as a gift or as part of payment, or bartered. Second, ideally only food that was actually consumed, as opposed to food purchases or total home-produced food, must enter in the consumption aggregate. HIES provides a detailed account of all transactions for each food item and also information on initial and final stocks, therefore an exact measure of actual consumption can be calculated. Third, non-purchased food items need to be valued and included in the welfare measure. HIES collects expenditures and quantities just for food purchases, whereas for all other transactions only quantities are recorded. Instead of collecting reference prices to value this consumption, unit values (expenditures divided by quantities) from purchases were calculated and used to estimate the monetary value of non-purchased items. Most food items are disaggregated enough to be regarded as relatively homogeneous within each category, however unit values are not prices, they will also reflect differences in the quality of the good. To minimize this effect, and to consider spatial and seasonal differences too, median unit values were computed at several levels: by household, cluster, aimag, strata and quarter. Hence if a household purchased a food item, the same unit value would be used to value its self-produced and in-kind consumption. If the household did not make any purchase but consumed a food item, unit values from the immediate upper level were used to estimate the value of consumption.
As in the case of food, non-food consumption is a simple and straightforward calculation. Again, all possible sources of consumption must be included and normalized to a common reference period. This component draws on data from both HIES and LSMS. As it was mentioned before, HIES collects information based on a diary kept by the households during 3 months. Data on an extensive range of non-food items is available, 242 items arranged in 14 different groups: clothing and footwear for men, women and children, jewelry and souvenirs, clothing materials, education, health, recreation, beauty and toilet articles and services, cultural expenses, household goods, durable goods, housing expenditures, transportation, and communication. Even tough most non-food items are too heterogeneous to try to calculate unit values, HIES does gather data on expenditures and quantities for most of them, yet only expenses were taken into account for the estimation of consumption. LSMS records information on education, health, rent of the dwelling, durable goods and energy expenses, using mostly a last year recall period. With the exception of durable goods, housing and energy, which will be dealt with later, this section covers the consumption of all the other non-food items.
Practical difficulties arise often for two reasons: the choice of items to include and the selection of the recall period. Regarding the first issue, the rule of thumb is that only items that contribute to the consumption are to be included. For instance, clothing, footwear, beauty articles and recreation are included. Others such as taxes are commonly excluded because they are not linked to higher levels of consumption, households paying more taxes are not likely to receive better public services. Capital transactions like purchases of financial assets, debt and interest payments should also be excluded. The case for lumpy or infrequent expenditures like marriages, dowries, births and funerals is more difficult. Given their sporadic nature, the ideal approach would be to spread these expenses over the years and thus smooth them out, otherwise the true level of welfare of the household will probably be overestimated. Lack of information prevents us to do that, so they are left out from the estimation. Finally, remittances given to other households are better excluded. The rationale for this is to avoid double counting because these transfers almost certainly are already reflected in the consumption of the recipients. Hence including them would increase artificially living standards.
Two non-food categories deserve special attention: education and health. In the case of education there are three issues to consider. First, some argue that if education is an investment, it should be treated as savings and not as consumption. Benefits from attending school are distributed not simply during the school period but during all years after. Second, there are life-cycle considerations, educational expenses are concentrated in a particular time of a person's life. Say that we compare two individuals that will pay the same for their education but one is still studying while the other finished several years ago. The current student might seem as better-off but that result is just related to age and not to true differences in welfare levels. One way out would be to smooth these expenses over the whole life period. Third, we must consider the coverage in the supply of public education. If all population can benefit from free or heavily subsidized education (as it happens in Mongolia) and the decision of studying in private schools is driven by quality factors, differences in expenditures can be associated with differences in welfare levels and the case for their inclusion is stronger. Standard practice was followed and educational expenses were included in the consumption aggregate. Excluding them would make no distinction between two households with children in school age, but only one being able to send them to school.
Health expenses share some of the features presented for education. Expenditures on preventive health care could be considered as investments. Differences in access to publicly provided services may distort comparisons across households. If some sectors of the population have access to free or significantly subsidized health services, whereas others have to rely on private services, differences in expenditures do not correspond to differences in welfare. But there are other factors to take into account. First, health expenditures are habitually infrequent and lumpy over the reference period. Second, health may be seen as a “regrettable necessity”, i.e. by considering in the welfare indicator the expenditures incurred by a household member that was sick, the welfare of that household is increased when in fact the opposite has happened. Third, health insurance can also distort comparisons. Insured households may register small expenditures when some member has fallen sick, while uninsured ones bigger amounts. We decided to include health expenses. As with education, excluding them would imply making no distinction between two households, both facing the same health problems, but only one paying for treatment. Besides, a positive relationship was found between health expenses and the rest of the consumption aggregate.
The second difficulty regarding non-food consumption is related with the election of the recall period. The key aspect to consider is the relationship between recall periods and frequency of purchases. Many non-food items are not purchased frequently enough to justify a weekly or monthly recall period, exceptions being for instance toiletries, beauty articles and payment of utilities. Generally recall periods are the last quarter or the last year. For most of non-food categories information comes only from HIES, thus just one option can be used, data based on a 3-month period, or in other words, a quarter. Still, a few non-food categories are available from both HIES and LSMS: mainly education and health. Aside from the fact of different recall periods, the other significant difference to keep in mind is that, for those two expenses, HIES collects expenditures at the household level, while the LSMS at the individual level. When the reference is the household, questions are normally more aggregated than when the same topics are asked to each household member. Generally households are known to provide a more accurate account of expenses when they are asked in more detail, which would favor the use of the LSMS modules. That is indeed the case of health expenses, where LSMS records a higher level than that of HIES. For education though, expenditures are very similar. Since the LSMS modules might capture better the long-term welfare of the household, it was decided to use them.
Ownership of durable goods could be an important component of the welfare of the households. Given that these goods last typically for many years, the expenditure on purchases is not the proper indicator to consider. The right measure to estimate, for consumption purposes, is the stream of services that households derive from all durable goods in their possession over the relevant reference period. This flow of utility is unobservable but it can be assumed to be proportional to the value of the good. A usual procedure involves calculating depreciation rates for each type of good based on their current value and age, which in this case is provided by the LSMS along with the number of durables owned by the household.
The estimation of this component involved three steps. First, a selection of durable goods was done. The LSMS supplies data on 47 durable goods, ranging from home appliances to furniture. However, a third of them were excluded because they were goods used for household businesses or fell under jewelry, dwelling or “other” categories. Second, to calculate implicit depreciation rates a non-linear regression for each of the remaining goods was run with the current unit value as the dependent variable on a constant and the age of the durable. This technique allows also for the possibility of applying multiple depreciation rates, for instance a higher one when the durable good is new. Finally, the stream of consumption is computed by multiplying the current value of the good times its depreciation rate, and aggregating these consumptions by household.
Housing conditions are considered an essential part of people's living standards. Nonetheless, in most developing countries limited or nonexistent housing rental markets pose a difficult challenge for the estimation and inclusion of this component in the consumption aggregate. As in the case of durable goods, the objective is to try to measure the flow of services received by the household from occupying its dwelling. When a household rents its dwelling, and provided rental markets function well, that value would be the actual rent paid. In Mongolia, housing value for non-renters households cannot be determined based upon on information from renters because very few cases reported renting their dwellings. Yet the LSMS asked households for estimates of how much their dwelling could be rented for and also how much their dwelling could be sold for. The implicit rental value can in principle be used in the consumption aggregate whenever actual rents are not reported. Implicit rents are a hypothetical concept though and the estimates may not always be credible or usable. An additional complication is that almost half of the population lives in gers, for which establishing a rent value appears to be even more difficult.
Two sets of hedonic housing regressions were run, one with the imputed rent value as the dependent variable and the other with the imputed value of the dwelling. The set of independent variables included characteristics of the dwelling such as main type of floor, walls, roof, number of rooms, access to water, electricity, heating, location, etc. This exercise was conducted separately for gers, detached houses and apartments. Results show that the value of the dwelling has a more consistent correlation with its characteristics and this is intuitively explained by the fact that even though households do not rent dwellings, they do buy and build them, so they report more accurately the overall value of the dwelling rather than a hypothetical rent. A second factor that favors the use of the property value is its higher response rate (more than 90% of the households reported this value compared to around 55% reporting imputed rents), which would suggest, as it was mentioned before, that households do have a better sense of the property value of their dwellings. However, the use of property values requires an additional assumption to arrive to an estimation of the services provided from housing and that is the depreciation rate of the dwelling. It was assumed that the annual rates were 3% for houses and apartments, and 6% for gers, in other words, houses and apartments will fully depreciate after 33 years and gers after 17 years. Two alternative sets of depreciation rates (2 and 5%, and 4 and 7%) produced very similar poverty measures. Therefore for the consumption aggregate, we used the estimated imputed rents derived from the imputed property values as estimates for the flow of services from housing, and otherwise actual rents if available.
The final non-food component that justified special attention was energy, meaning basically expenditures on heating and electricity. Mongolia is a country that endures extreme weather conditions, during winter temperatures can easily reach -40 degrees Celsius and in the summer 30 plus degrees. While summer may pose fewer inconveniences, winter is indeed a serious matter. Winters are long, they last on average 6 months and with usual below zero temperatures. For instance, average temperatures in January and February in the capital are -25C. This means that heating becomes a basic and essential necessity for households all over the country, and in some cases it could be a very significant and important component of their consumption.
Both surveys provide information on energy but the LSMS is the one that contains a very comprehensive and detailed module, hence it is likely to be much more accurate than the corresponding HIES section. Electricity and lighting expenses offered no problems for their inclusion in the welfare indicator. Heating was a different case. Heating is provided to households from either central or local systems or simple heating units fueled by firewood, coal or dung. While information on the former was appropriately captured, the latter presented a few complications. The questionnaire collected data on average purchases (expenditures and quantities) and collection (quantities) per winter and non-winter month for those three main sources of fuel. First, to value consumption coming from collected fuel, unit values for each one of the three main fuels were applied to their respective collected quantities. In urban areas, where most fuel is purchased, unit values were estimated from actual purchases recorded in the LSMS following a similar procedure as in the case of valuing food collection. In rural areas tough, where most fuel is collected and there is no market for fuel, the same method will likely overestimate the value of consumption (Since no transactions are registered at the cluster level and very few at the aimag level, unit values are probably drawn from urban areas). Information on household fuel consumption was gathered from several aimag statistical offices and unit values were obtained from there. Second, given that the recall period was the last year, we needed to make an assumption on the duration of winter and non-winter seasons in order to arrive to a monthly figure. It was assumed that each season lasts on average 6 months.
3. Price adjustment
Mongolia shows remarkable seasonal price differences, especially for food items, with prices in the spring (April to June) commonly 10% higher than in autumn. At the same time, across all seasons there are also regional price differences. In particular in Ulaanbaatar, prices are relatively higher than in the rest of the country. Therefore, in order to properly measure living standards, expenditure values need to be corrected for such differences using some price indexes. A price index is made of two components: prices and budget shares that attach the proper weights to prices. It follows that differences of price indices can come both from prices and consumption patterns.
The household survey provides information on budget shares as well as information on implicit prices (unit values) paid by the household. Moreover, together with the household survey the NSO also conducted a price questionnaire in soum centers collecting information on about 250 prices, and regularly collects prices for about 140 items in all aimag centers. All this provides a rich source of information, which was used to construct a Paasche price index at the cluster level. In each cluster generally between 8 and 10 households have been interviewed and prices they face as well as consumption patterns tend to be very similar.
Budget shares were computed from the household surveys, as well as food prices. However, it is important to note that the household survey does not collect information on prices themselves, but on implicit prices, obtained dividing expenditure by quantities purchased. Inevitably, implicit prices represent also differences in quality of the item purchased. Quality differences are generally considered acceptable for food items, but are more problematic for non-food items, which are likely to be less homogenous in nature (also questions on non-food items are less detailed than those for food ones). On the contrary, both the soum and aimag centers questionnaires collected information on actual prices and on much more well defined items. Nonetheless, the soum center price questionnaire was not always of the desired quality, some of the items show price differences that are too large, suggesting that in such cases prices of items of rather different quality were collected. This is to be expected in fragmented and incomplete markets, where the enumerator might have been compelled to substitute items that were not found.
Instead, the aimag centers prices appear to be more accurate because they are the result of a permanent activity, prices are collected in the same outlets and with more precise guidelines about the type of item for which the price is sought. Both for the soum and the aimag price questionnaire, information is not available for each household, but is representative respectively for the soum or aimag. However, it is likely that both within the same soum, and indeed the same aimag, prices of non-food items show a relatively small variation. This is because price differences for these items are mainly due to transportation costs (from Ulaanbaatar), and the soum/aimag price already captures most of such costs.
More problematic is the fact that while for food items budget shares are immediately matched with 'prices', when information on prices is taken from the price questionnaires, the correspondent budget share needs to be properly identified, and in some cases, where such correspondence does not exist, key items are considered to be representative for the budget shares of similar items. For instance, in the case of transportation expenditure, the only price that was used was the one of petrol (petroleum A-76).
The index confirms that living costs in Ulaanbaatar are higher than anywhere else in the country and it also shows the seasonality effects: the index is higher in the first and second quarters and then decreases in the following quarters.
4. Household composition adjustment
The final step in constructing the welfare indicator involves going from a measure of standard of living defined at the household level to another at the individual level. Ultimately the concern is to make comparisons across individuals not households. Consumption data are collected typically at the household level (usual exceptions are health and education expenses), so computing an individual welfare measure generally is done by adjusting total household consumption by the number of people in the household, and assigning that value to each household member. Common practice to do this is to assume that all members share an equal fraction of household consumption, however as it will be explained later that is a very particular case.
Two types of adjustments have to be made to correct for differences in composition and size. The first relates to demographic composition. Household members have different needs based mainly on their age and gender, although other characteristics can also be considered. Equivalence scales are the factors that reflect those differences and are used to convert all household members into “equivalent adults”. For instance, children are thought to need a fraction of what adults require, thus if a comparison is made between two households with the same total consumption and equal number of members, but one of them has children while the other is comprised entirely by adults, it would be expected that the former will have a higher individual welfare than the latter. Unfortunately there is no agreement on a consistent methodology to calculate these scales. Some are based on nutritional grounds, a child may need only 50% of the food requirements of an adult, but is not clear why the same scale should be carried over non-food items. It may very well be the case that the same child requires more in education expenses or clothing. Others are based on empirical studies of household consumption behavior, although with more analytical grounds, they do not command complete support either.
The second adjustment focuses in the economies of scale in consumption within the household. The motivation for this is the fact that some of the goods and services consumed by the household have characteristics of “public goods”. A good is said to be public when its consumption by a member of the household does not necessarily prevent another member to consume it too. Examples of these goods could be housing and durable goods. For example, one member watching television does not preclude another for watching too. Larger households may spend less to be as well-off as smaller ones. Hence, the bigger the share in total consumption of public goods, the larger the scope for economies of scale. On the other hand, private goods cannot be shared among members, once they have been consumed by one member, no other can. Food is the classic example of a private good. It is often pointed out that in poor economies, food represents a sizeable share of the household budget and therefore in those cases there is little room for economies of scale.
Both adjustments can be implemented using the following approach:
AE = (A + aK)q
where AE is the number of adult equivalents of the household, A is the number of adults, K the number of children, a is the parameter that measures the relative cost of a child compared to an adult and q represents the extent of the economies of scale. Both parameters can take values between zero and one. It is been reported that in developing countries, children are relatively cheaper than adults, perhaps with values of a as low as 0.3 while in developed ones values are closer to one. At the same time, in poorer economies food is often the most important good in the household consumption, and given that is a private good, the budget share of public goods is limited and so is the scope for economies of scale, perhaps with q close to 1, whereas in richer countries around 0.75.
It was mentioned that standard practice is to use a per capita adjustment for household composition and that is also followed here. This is a special case of the above formulation, it happens when a and q are set equal to one, so all children are treated as if their cost relative to adults were the same and there is no room for economies of scale. In other words, all members within the household consume equal shares of the total consumption and costs increase in proportion to the number of people in the household. In general, per capita measures will underestimate the welfare of households with children as well as larger households with respect to families with no kids or with a small number of members respectively. It is important then to conduct sensitivity analysis to see how robust the poverty measures and rankings are to different assumptions regarding child costs and economies of scale.
5. The poverty line
The poverty line can be defined as the monetary cost to a given person, at a given place and time, of a reference level of welfare (Ravallion, 1998). If a person does not attain that minimum level of standard of living, she will be considered as poor. But setting poverty lines could be a very controversial issue because not only people disagree on what “minimum” is but also of its eventual effects on monitoring poverty and policy making decisions.
The poverty line will be absolute because it fixes a given welfare level, or standard of living, over the domain of analysis. This guarantees that comparisons over time or across individuals will be consistent e.g. two persons with the same welfare level will be treated the same way regardless of the location where they live. Second, the reference utility level is anchored to certain attainments, generally nutritional ones, for instance, obtaining the necessary calories to have a healthy and active life. Finally, the poverty line will be set as the minimum cost of achieving that requirement.
The Cost of Basic Needs method was employed to estimate the nutrition-based poverty line. This approach calculates the cost of obtaining a consumption bundle believed to be adequate for basic consumption needs. If a person cannot afford the cost of the basket, it will be considered to be poor. First, it shall be kept in mind that the poverty status focuses on whether the person has the means to acquire the consumption bundle and not on whether its actual consumption met those requirements. Second, nutritional references are used to set the utility level but nutritional status is not the welfare indicator. Otherwise, it will suffice to calculate caloric intakes and no costing would be necessary. Third, the consumption basket can be set normatively or to reflect prevailing consumption patterns. The latter is undoubtedly a better alternative. Lastly, the poverty line comprises two main components: food and non-food.
The first step in setting this component is to determine the nutritional requirements deemed to be appropriate for being healthy and able to participate in society. Clearly, it is rather difficult to arrive to a consensus on what could be considered as a healthy and active life, and hence to assign caloric requirements. Common practice is to establish 2,100 calories per person per day as the reference for energy intake. Second, a food bundle must be chosen. In theory, infinite food bundles can provide that amount of calories. One way out of this is to take into consideration the existing food consumption patterns of a reference group in the country. It was decided to use the bottom 40% of the population, ranked in terms of real per capita consumption, and obtain its average consumed food bundle. It is better to try to capture the consumption pattern of the population located in the low end of the welfare distribution because it will probably reflect better the preferences of the poor. Hence the reference group can be seen as a first guess for the poverty incidence. Third, caloric conversion factors were used to transform the food bundle into calories. The main source for these factors was the Food Research Center, which is a unit of the Ministry of Health of Mongolia. Alcohol, tobacco and meals eaten outside the household were excluded from this calculation, the former because they can be regarded as non-essential and the latter because it is very difficult to approximate caloric intakes for them. For all of the remaining food items, it was possible to assign a caloric factor. Fourth, median unit values were derived in order to price the food bundle. Unit values were computed using only transactions from the reference group. Again, this will capture more accurately the prices faced by the poor. Fifth, the average caloric intake of the food bundle was estimated, so the value of the food bundle could be scaled proportionately to achieve 2,100 calories per person per day. For instance, the average daily caloric intake of the bottom 40% of the population in Mongolia was around 1,345 calories per person and the daily value of the food bundle was Tugrug 307 per person. Hence the value of the daily poverty line is Tugrug 480 ( = Tugrug 307 x 2,100 / 1,345 ) per person. Table B.3 shows the caloric contribution of the main food categories as well as the their respective share in the cost of the food poverty line.
Setting this component of the poverty line is far from being a straightforward procedure. There is considerable disagreement on what sort of items should be included in the non-food share of the poverty line. However, it is possible to link this component with the normative judgment involved when choosing the food component. Being healthy and able to participate in society requires spending on shelter, clothing, health care, recreation, etc. A usual practice is to scale up the food poverty line to allow for basic non-food items, which can be done by dividing the food poverty line by some estimation of the budget share devoted to food. The advantage of this is that the non-food component can be based on the prevailing consumption behavior of a reference group and no pre-determined non-food bundle is needed.
The initial step is to choose a reference group. There are two ways in which this is usually done. The first is to determine the food share of the population whose food expenditures are equal to the food poverty line. The rationale behind is that if an individual spends in food what was considered appropriate for being healthy and maintaining certain activity levels, it can be assumed that this person has also acquired the necessary non-food items to support its lifestyle. The resulting poverty line is called the upper or higher poverty line. The second way to calculate the food share is to estimate it from the population whose total expenditures are equal to the food poverty line. The justification is that these people have substituted basic food needs in order to satisfy some non-food needs, therefore that amount can be interpreted as the minimum necessary allowance for non-food spending.
Two different procedures to calculate the non-food component can be proposed. One relies on econometric techniques to estimate the Engel curve, e.g. the relationship between food spending and total expenditures. Another is to use a simple non-parametric calculation as suggested in Ravallion (1998). The advantages of the latter is that no assumptions are made on the functional form of the Engel curve and that weights decline linearly around the food poverty line i.e. the closer is the household to the food poverty line, the higher its weight. This procedure was used to determine the non-food components for the upper and lower poverty lines. For instance, in the case of the upper poverty line, first food shares are estimated from those households whose food expenditures lie within plus and minus one percent around the poverty line. The same exercise is then repeated for households lying plus and minus two percent, three percent, and up to ten percent. Second, these ten mean food shares are averaged and that will be the final food share of the poverty line. Finally, the non-food component can be easily estimated.
6. Poverty measures
Even though there is an extensive literature on poverty measurement, attention will be given to the class of poverty measures proposed by Foster, Greer and Thorbecke (1984).
The headcount index (a=0) gives the share of the poor in the total population, i.e. it measures the percentage of population whose consumption is below the poverty line. This is the most widely used poverty measure mainly because it is very simple to understand and easy to interpret. However, it has some limitations. It takes into account neither how close or far the consumption levels of the poor are with respect to the poverty line nor the distribution among the poor. The poverty gap (a=1) is the average consumption shortfall of the population relative to the poverty line. Since the greater the shortfall, the higher the gap, this measure overcomes the first limitation of the headcount. Finally, the severity of poverty (a=2) is sensitive to the distribution of consumption among the poor, transfers among the poor will leave unaffected the headcount or the poverty gap but will increase this measure. It applies a relatively higher weight to the largest poverty gaps.
These measures satisfy some convenient properties. First, they are able to combine individual indicators of welfare into aggregated measures of poverty. Second, they are additive in the sense that the aggregate poverty level is equal to the population-weighted sum of the poverty levels of all subgroups of the population. Third, the poverty gap and the severity of poverty satisfy the monotonicity axiom, which states that even if the number of the poor is the same, but there is a welfare reduction in a poor household, the measure of poverty should increase. And fourth, the severity of poverty will also comply with the transfer axiom: it is not only the average welfare of the poor that influences the level of poverty, but also its distribution. In particular, if there is a transfer from one poor household to a richer household, the degree of poverty should increase.
Finally, in the final report all poverty measures are shown with their respective standard errors. Since those estimations are based on surveys and not on census data, standard errors must reflect the elements of the sample design i.e. stratification and clustering. Ignoring them will risk, when carrying out poverty comparisons, mixing up true population differences with differences in sampling procedures.