# Introduction ealth has traditionally and commonly been measured using monetary indicators such as income and consumption (Hargreaves et al., 2007). Income is "the amount of money received during a period of time in exchange for labour or services, from the sale of goods or property, or as a profit from financial investments" (O' Donnell et al., 2008;70). On the other hand, consumption is "the final use of goods and services, excluding the intermediate use of some goods and services in the production of others" (pp, 70). While there could be some differences in defining these two concepts, the approach to use them as welfare indicators has resulted in the production of social protection policies in various countries including Botswana. However, some researchers have debated the adequacy of the two monetary indicators in capturing status of welfare; hence alternative approaches have been proposed to serve this purpose. It has been observed that despite the findings of assets being the underlying determinants of poverty in the developing world, little attention (safe for human capital proxied by education) is given to them, resulting in the objectives to address only income (and/or expenditure) poverty (Sahn and Stifel, 2003). The use of assets as a welfare indicator has however, not escaped criticism. Some argue that ownership does not capture the issue of assets quality (Falkingham and Namazie, 2002). Thus, the process of collecting data on assets may not differentiate households that own new or old assets, cheap or expensive ones etc. Notwithstanding that, the authors argue that in a number of countries, such traits would not change the overall picture of wealth. Filmer and Scott (2008) make references to the extensive use of asset indices in previous studies. The authors indicate that this index has been used for analysis of poverty change, inequality (in health and education outcomes), and for program targeting and evaluation. While this pattern is observed in the literature, little (or no) evidence exists in Botswana for utilizing assets to inform welfare status. This is despite that the surveys conducted and the previous census collected data on assets. This paper therefore fills this gap. The paper n mn m m m n n Y Y Y Y PCm Y Y Y Y PC ? ? ? ? ? ? ? ? + + + + = + + + + = ... . . ... 3 3 2 2 1 1 1 3 13 2 12 1 11 1 In the above specification, mn ? is the weight for the m th Principal Component (PC) and the n th variable, given set of variables from Y 1 to Y n . The weights of the Results of this paper are important as they may assist policy makers to identify areas of concern to uplift household wealth, which should facilitate not only the attainment of MDGs but also the country's Vision 2016 aspirations. The rest of the paper is organised as follows. Section II discusses the methodology while section III discusses data source and descriptives. Results are presented and discussed in section IV, and section V concludes. # II. # Methodology a) Computation of an Index The use of asset/welfare index is common in situations where data on either income or consumption was not collected. This approach is therefore relevant for this paper, with the 2011 population and housing census, which only asked about the source of income. Moreover, "the index captures a dimension of economic status" (Filmer and Scott, 2008;4) and gives more reflection on long run household wealth (Filmer and Pritchett, 2001). Some of the issues to be considered in computing the index include choice of assets and their weights. Several approaches to computing the index exist. One of them is the simple total sum of assets from a dummy variable of whether a particular household owns assets or not (Case, Paxson and Ableidinger, 2004;Montgomery et al., 2000). This approach has been termed an "arbitrary approach" as it assumes equal weights for the different assets (O' Donnell et al., 2008;Vyas and Kumaranayake, 2006). Another approach is the use of statistical techniques which address the issues of weights in the index. The two commonly used techniques are the factor analysis and Principal Component Analysis (PCA). In this paper we computed the wealth index from a technique of PCA, which is a tool used to reduce a number of variables into one. It is mathematically specified as follows: PCs are represented by the eigenvectors of the correlation matrix. However, if the data is standardized the eigenvectors would be of the co-variance matrix. On the other hand, the variance of the PCs is given by the eigenvalues (Vyas and Kumaranayake, 2006). In the output, components are ordered according to their proportion of variation that they explain in the original data; with those in the top positions explaining larger amounts of variation. The index was computed from housing conditions (type of houses, wall, floor, and roof materials), living conditions (water source, toilet facility and energy sources for lighting, cooking and heating) as well as ownership of durable assets (Television, radio, sewing machines, watch etc). While there is no defined criteria for the choice of assets (Montgomery et al., 2000); ours was influenced by the bearing that the variables might have on the Millennium Development Goals. For instance, source of water, sanitation and flooring material affect hygiene. Source of energy for cooking may affect the environment and respiratory diseases that cause deaths. Some of the variables were in categorical form, which is not suitable for the PCA technique and were therefore converted to binary variables. After computing the wealth index, households were then classified into quintiles. The decision to choose five groups (quintiles) was among others informed by previous empirical work. According to literature, the commonly used cut-off points are classification into quintiles (Gwatkin et al. 2000;Filmer and Pritchett 2001). This is done to differentiate households into socio economic categories; to show wealth status within a population. We used SPSS (Version 18) for analysis. # III. # Data Source and Descriptives The pattern for type of housing unit is dominated by detached houses (43%) followed by rooms and traditional house with 23 percent and 13 percent, respectively. Other types (town house, mixed, flat, shacks and movable) accounted for a share of 10 percent or less. Majority (82%) of households had their walls made out of conventional bricks/blocks while the remaining shares were distributed amongst corrugated iron, asbestos, wood, stones and poles and reeds. A larger proportion (65%) had cement as a floor material, 22 percent with floor tiles and 0.07 percent with brick/stone. Roof material is dominated by corrugated iron (74%), followed by roof tiles (13%), while the least share was for concrete (0.3%). Regarding water supply, majority (40%) of households had piped outdoors while 30 percent had piped indoors. Thus, majority appear to be accessing water from improved sources. This pattern was also observed by previous studies (Statistics Botswana, 2011). About 15 percent of households sourced water from communal taps. Other water sources including bouser/tanker, well, borehole, and dam/pan had a share of less than 10 percent. Those who owned flush toilet accounted for a share of about 25 percent followed by those who owned pit latrines with 24 percent. However, 18 percent of households shared pit latrines, 5 percent used neighbor`s pit latrines, and 9 percent shared flush toilet. While there is dominance of use of pit latrines, it is promising that the use of flush toilets (whether owned or shared) is also visible. The shares for those who used communal toilet facilities were less than a percent. The above presents a hopeful trend towards the achievement of the Millennium Development Goal 7 of ensuring environmental sustainability. More than half of households used electricity as a principal source of energy for lighting while 30 and 11 percent used paraffin and candles respectively. About 41 percent of households used wood as a source of energy for cooking followed by 38 percent who used gas. The use of wood also dominated sources of energy for heating (48%), followed by electricity with a share of 17 percent. About 15 percent of households owned van/bakkie; 2 percent owned tractors and 20 percent owned cars. The shares of ownership status for donkey carts and bicycles stood at 12 and 10 percent respectively, while motor bike and boat were each owned by about a percent of households. About 43 percent owned the refrigerator and 5 percent owned sewing machine. Given that these assets have a positive factor score, their ownership implies the likelihood of improved welfare for households. On the other hand, majority (90%) owned cell phones while 11 percent had telephones (landlines). About 61 percent owned radios and 54 percent owned televisions. This pattern presents a positive outcome towards an "informed nation" as these assets are among the primary sources of information. IV. # Results and Discussions We begin by presenting the welfare status by census district (Table 1 The paper used data from the 2011 population and housing census, which had 550944 households. Table A1 in the annex presents descriptive statistics. The fourth column of Table A1 shows the factor score, which is basically the first principal component (weight), used to create a household score (Houweling et al., 2003). A positive score suggests that a variable is associated with a higher economic status (wealth) while the opposite is true for a negative score. Thus, from Table A1, with regard to the type of housing unit, traditional, mixed, movable, shacks and rooms will be associated with lower economic status. The use of mud bricks/blocks or poles and reeds for floor would also reduce household wealth. of households with better status of wealth. The proportions of households increase as we move from the lower (poorest) wealth status to the higher (richest) status. For instance, 0.6 (1.2) percent of households are in the poorest wealth status in Gaborone (Francistown) compared to 45 and 29 percent in the richest status respectively. This pattern is also observed in Lobatse, Selebi Phikwe, Sowa Town and Jwaneng, with some minor variations. These results corroborate findings from previous studies, that these districts had lower poverty incidence compared to others (CSO, 2008; Statistics Botswana, 2013). For instance in 2002/03 poverty incidence stood at 0.076, 0.159, and 0.018 percent for Gaborone, Francistown and Orapa respectively. The districts of Ngamiland West, Kweneng West, Ngwaketse West, CKGR, and Ghanzi had the highest proportions of households in the poorest status (all over 40%). These results are consistent with those of previous survey by Statistics Botswana (2013) where 1 suggests that female headed households are better off. This pattern is observed up to the fourth category of welfare. About 22 percent of male headed households are in the poorest status of wealth compared to 18 percent of female headed households. However, in the richest category we observe higher proportion of male headed households than that of female headed households. While this is the case, it is also evident that from the second to the richest status of wealth the proportions of female headed households declined while that of male headed households increased. poverty rates were found to be higher in such districts. Ngwaketse, Ngwaketse West, Mahalapye, Bobonong, Tutume, Ngamiland and Kgalagadi are generally characterized by larger proportions of households in the poorer status of wealth than those in the richer status. For instance, about 49 percent of households in Ngwaketse West are in the poorest status of wealth compared to 7 percent of those in the richest status; while 29 percent of households in Kgalagadi North are in the lower wealth status compared to 13 percent for those in a richer state. We conclude that generally the urban (or city/town) districts are characterized by better wealth status than their rural counterparts. One of the possible explanations for the observed pattern could be employment opportunities found in urban areas and cities/towns. Although there are various modes of assets acquisition (including inheritance), income from employment is likely to improve status of asset ownership. Table 2 presents the share of wealth status by marital status of heads of households. Among households with married heads, a higher proportion (25.6%) is in the richest category of wealth followed by those in the fourth category (20.7%). The least share of households whose heads are married is accounted for by those in the poorest status of wealth. This may suggest that being married is likely to improve the household status of wealth. Similarly, households whose heads were never married are more concentrated in the richest category than in the poorest category. This may not be surprising given that pervious studies found a comparable poverty incidence in households with married and never married heads (BIDPA, 2010). # Level There are higher proportions (in the poorest category) of households whose heads are separated, living together and widowed. As seen in Table 2, 24 percent of households whose couples are living together are in the poorest category of wealth compared to 16 percent of those in the richest category. About 30 percent of households headed by separated heads are in the poorest category compared to 14 percent in the richest category. As for widowed households, the proportions are 24 and 12 percent for poorest and richest categories respectively. The pattern for households with divorced heads is interestingly similar to that of households with married and never married heads, safe for the third category of wealth status. This could be argued to be against the expectations as divorce may result in a reduced status of assets ownership. are more concentrated in the better status of wealth. In fact the proportions in both the poorest and richest categories are a mirror image of the pattern observed in households with uneducated heads. This could suggest that education might be a determinant of households' wealth status; it may improve acquisition of assets to better the status of household wealth. Household Wealth Status in Botswana: An Asset Based Approach Table 3 : Share (%) of Wealth Status by Education Status of Household Heads those whose heads had no education. On the other hand, households whose heads had tertiary education Table 3 presents the pattern for wealth status by level of education attained by households' heads. As evident in the table, the status of wealth is positively related to the level of education of the household head. For instance, about 7 percent of households headed by those who have never been to school are in the richest category of wealth compared to about 40 percent in the poorest category. A similar pattern is observed for households whose heads had primary and secondary education, who however appear to be faring better than # Conclusions This paper assessed welfare status using the index computed from the technique of Principal Component Analysis. To our knowledge this approach has not been done in Botswana. Therefore, it may not be easy to conclusively note whether there has been an improvement or not, in addition to what has been done so far. Therefore this paper may be seen as the baseline against which future progress will be tracked. Results have shown that generally there is better status of wealth among urban districts, female headed households as well as in households with married heads. Further, education also appears to be an important determinant of asset acquisition. Results revealed a positive relation between wealth status and educational level of heads of households. Results from our analysis suggest that from a policy point of view, there is need to broaden issues of consideration in designing programmes for poverty eradication. Thus, there is need to also focus on economic and social forces that contribute to assets inequality, given that sometimes both the policies and programmes for poverty eradication would be based on individuals' ability to accumulate productive assets. Moreover, the problem of income inequality might be exacerbated by unequal distribution of income generating assets, hence the need for consideration of assets. Although some reports suggest that Botswana is on track to meeting MDG 1 of halving extreme poverty and hunger, such needs to be supplemented by consideration of assets with the view to try to address the multidimensionality of poverty, especially that the target may be seen to have been narrowed to "income' or expenditure as welfare measures. 1![Figure 1 : Share(%) of Wealth Status by Gender of Household Heads](image-2.png "Figure 1 :") 1DistrictPoorestSecondMiddleFourthRichestGaborone448 (0.6)8692 (11.6)15049(20.1)17019 (22.7)33749(45.0)Francistown384(1.2)5153(16.5)7333(23.4)9501(30.4)8926(28.5)Lobatse200(2.2)1898(20.6)2438(26.5)2012(21.8)2666(28.9)Selebi Phikwe281(1.7)2851(17.8)3347(20.8)5097(31.7)4483(27.9)Orapa0(0.0)1(0.0)62(1.9)732(22.22497(75.9)Jwaneng449(7.6)281(4.7)1063(17.9)1400(23.6)2747(46.2)Sowa Town28(2.4)44(3.7)42(3.5)534(44.8)543(45.6)Ngwaketse7551(24.0)8503 (27.0)5947(18.9)5841(18.6)3639(11.6)Barolong3300(24.0)5146(37.4)2389(17.4)1614(11.7)1309(9.5)Ngwaketse West1725(48.5)999(28.1)328(9.2)264(7.4)240(6.7)South East952(4.0)2894(12.1)5689(23.7)7519(31.3)6936(28.9)Kweneng East8488(12.4)14158(20.7)17961(26.3)17128(25.2)10504(15.4)Kweneng West6948(56.8)2524(20.6)907(7.4)751(6.1)11012(9.0)Kgatleng3427(13.8)5866(23.5)5474(22.0)5622(22.6)4528(18.2)Serowe/Palapye12508(27.1)9953(21.5)8974(19.4)8234(17.8)6519(14.1)Mahalapye8731(29.3)8227(27.6)5217(17.5)4265(14.3)3359(11.3)Bobonong6186(32.3)5025(26.2)3607(18.8)2544(13.3)1794(9.4)Boteti5879(41.7)2309(16.4)2527(17.9)2114(15.0)1281(9.1)Tutume14764(38.5)9064(23.6)6658(17.4)4621(12.0)3246(8.5)North East3001(18.9)4476(28.2)3446(21.7)2800(17.6)2142(13.5)Ngamiland East6262(28.8)3806(17.5)4648(21.4)4263(19.6)2758(12.7)Ngamiland West8413(63.9)1888(14.3)1299(9.9)900(6.8)664(5.0)Chobe1142(16.7)1030(15.1)1675(24.5)1817(26.6)1166(17.1)Okavango Delta191(29.2)242(36.9)200(30.5)21(3.2)1(0.2)Ghanzi4636(40.8)1731(15.2)1626(14.3)1920(16.9)1442(12.7)CKGR10(47.6)0(0.0)1(4.8)2(9.5)8(38.1)Kgalagadi South2682(33.7)1967(24.7)1221(15.3)1076(13.5)1010(12.7)Kgalagadi North1607(29.0)1444(26.1)1073(19.4)682(12.3)736(13.3)Source: Author computed from 2011 population and housing census data setFigure 1 presents household wealth status bygender of the household heads. Comparatively, theoverall picture presented in Figure 2Marital StatusPoorestSecondThirdFourthRichestMarried17.118.418.220.725.6Never Married18.319.521.021.020.1Living Together24.320.820.418.815.7Separated29.821.918.715.414.3Divorced17.119.018.019.826.1Widowed23.724.921.117.912.4Source: Author Computed from 2011 population and housing census data set AShared pit latrine Laptop Shared dry compost Desktop0.1823 0.1123 Annex 0.0010 0.09630.3861 0.3157 0.0321 0.2949-0.039 0.421 -0.032 0.393Communal Flush0.00120.03400.00728 ( E )Variable Communal VIP Communal pit latrine Traditional Communal dry compost Mixed Neighbours` Flush Detached Neighbours`VIP Semi Detached Neighbours pit latrine Townhouse/terraced Neighbour`s compost Flats/apartments Part of commercial building Electricity Movable Shack Rooms Conventional Bricks/Blocks Petrol Diesel Solar power Gas Bio Gas Mud bricks/blocks Mud and Poles/Cow dung/thatch reeds Wood Paraffin Poles and reeds Candle Corrugated Iron/zinc Asbestos Electricity Wood Petrol Stone Diesel Solar Power Cement Gas Floor tiles Bio Gas Mud Wood Mud/dung Paraffin Wood Cow dung Brick/stone Coal None Crop Waste Slate Thatch Roof Tiles Corrugated Iron Asbestos Concrete Other Piped indoors Piped outdoors Neighbour`s tap Communal tap Bouser/tanker Well Borehole River/stream Dam/pan Rain water tank Spring Water Wheel barrow Motor Bike Mokoro/Boat Bicycle Petrol Diesel Solar Power Gas Bio Gas Wood Paraffin Cow dung Coal Charcoal Van/bakkie Tractor Car Donkey Cart Electricity CharcoalMean 0.0004 0.0060 Type of Housing Unit 0.1319 0.0006 0.1000 0.0013 0.4340 0.0020 0.0457 0.0462 0.0193 0.0003 0.0153 Energy for Lighting 0.0014 0.5324 0.0070 0.0167 0.2286 Wall Material 0.8150 0.0015 0.0077 0.0051 0.0028 0.0002 0.0871 0.0548 0.0356 0.3002 0.0100 0.1101 0.0216 Energy for Cooking 0.0028 0.1779 0.0040 0.0006 0.0005 0.0009 0.0008 Floor Material 0.6471 0.3789 0.2199 0.0092 0.0535 0.4119 0.0499 0.0167 0.0019 0.0007 0.0007 0.0004 0.0235 0.0002 0.0067 0.1113 0.1292 0.7352 0.0091 0.0028 0.0057 Water Supply 0.3020 0.3990 0.0564 0.1477 0.0114 0.0093 0.0491 0.0139 0.0072 0.0010 0.0005 0.3314 Toilet Facility 0.0062 0.0065 0.0989 0.0009 0.0003 0.0014 0.0102 0.0006 0.4766 0.0026 0.0005 0.0013 0.0015 Other Assets (durables) 0.1509 0.0197 0.1981 0.1170 0.1675 Energy for Heating Roof Material 0.0013Standard Deviation 0.0206 0.0769 0.3384 0.0249 0.3001 0.0355 0.4956 0.4460 0.2089 0.2100 0.1375 0.0162 0.1229 0.3789 0.4990 0.8351 0.1282 0.4199 0.3883 0.0388 0.0873 0.0709 0.0527 0.0146 0.2820 0.2276 0.1854 0.4583 0.996 0.3130 0.1455 0.0531 0.3824 0.0635 0.0252 0.0221 0.0300 0.0278 0.4779 0.4851 0.4142 0.0954 0.2250 0.4922 0.2177 0.1280 0.0437 0.0273 0.0261 0.0191 0.1516 0.0130 0.0815 0.3145 0.3354 0.4412 0.0951 0.0527 0.0755 0.4591 0.4897 0.2307 0.3548 0.1062 0.0958 0.2160 0.1171 0.0844 0.0316 0.0230 0.4707 0.0787 0.0802 0.2985 0.0303 0.0169 0.0369 0.1005 0.0236 0.4995 0.0506 0.0217 0.0367 0.0392 0.3579 0.1390 0.3986 0.3214 0.3735 0.0364-0.017 Score -0.060 -0.034 -0.618 -0.014 -0.175 -0.037 0.463 -0.212 0.176 -0.016 0.130 0.168 0.808 0.003 -0.071 -0.163 -0.039 0.000 -0.108 -0.015 0.007 -0.003 0.677 -0.442 -0.311 -0.522 -0.392 -0.296 -0.152 -0.171 0.457 0.004 0.001 -0.080 0.011 -0.019 0.010 0.427 -0.097 0.036 0.613 -0.768 -0.382 -0.062 -0.379 -0.013 -0.007 0.004 -0.016 0.010 -0.239 -0.560 0.429 0.060 0.090 0.039 -0.077 0.695 -0.004 -0.190 -0.417 -0.100 -0.143 -0.314 -0.172 -0.014 0.057 0.000 -0.014 -0.021 -0.007 -0.121 0.004 0.001 0.016 0.071 0.010 -0.680 -0.023 -0.008 0.008 0.021 0.298 0.073 0.482 -0.246 0.533 0.012 0.005Year 2015 -Global Journal of Human Social Science ( E )Own Flush Sewing Machine0.2524 0.04640.4349 0.21040.657 0.120Own VIP Refrigerator0.0183 0.43470.1339 0.4957-0.008 0.708Own pit latrine Cell phone0.2367 0.89730.4251 0.3036-0.141 0.406Own dry compost Telephone0.0028 0.10830.0526 0.3108-0.063 0.326Shared Flush Radio0.0860 0.61490.2803 0.48660.197 0.323Shared VIP Television0.0143 0.54090.1187 0.49830.005 0.723© 2015 Global Journals Inc. (US) © 2015 Global Journals Inc. (US) Household Wealth Status in Botswana: An Asset Based Approach * Review of the National Strategy for Poverty Reduction: Social Protection for poverty and vulnerability reduction 2010 Botswana Institute for Development Policy Analysis (BIDPA). Unpublished consultancy report, BIDPA * Orphans in Africa: Parental death, poverty, and school enrollment ACase CPaxson AbleidingerJ Demography 41 3 2004 * Botswana Census based poverty map report Central Statistics Office (CSO) Gaborone, Botswana Department of Printing and Publishing Services 2008 * Measuring health and poverty: a review of approaches to identifying the poor: London: Department of International Development Health Systems Resource Centre JFalkingham CNamazie 2002 * Estimating wealth effect without expenditure data-or tears; an application to educational enrollments in states of India DFilmer LHPritchett Demography 38 2001 * Assessing Asset Indices DFilmer KScott 2008 World Bank 4605 World Bank Policy Research Working Paper * Socio-Economic Differences in Health, Nutrition and Population. HNP/Poverty Thematic Group DRGwatkin SRustein JKiersten RPande AWagstaff 2000 The World Bank Washington DC * Assessing household wealth in health studies in developing countries: A comparison of participatory wealth ranking and survey techniques from rural South Africa JRHargreaves LAMorison JS SGear JCKim MBMakhubele JD HPorter CWatts PMPronyk Household Wealth Status in Botswana: An Asset Based Approach 8 2007 4 * Measuring health inequality among children in developing countries: Does the choice of indicator of economic status matter? THouweling AEKunst JPMackenbanch International Journal of equity in health 2 8 2003 * Measuring living standards with proxy variables MRMontgomery KGragnolati ABurke EParedes Demography 37 2000 * Analyzing health equity using household survey data: A guide to techniques and their implementation OO' Donnell EDoorslaer AWagstaff MLindelow 2008 World Bank Washington DC * Exploring alternative measures of welfare in the absence of expenditure data DESahn DStifel Review of Income and Wealth, Series 49 2003 * Botswana Core Welfare Indicator Survey Report Statistics 2013 Statistics Botswana * Constructing socio-economic status indices: how to use principal component analysis SVyas KumaranayakeL 2006 Oxford University