«2015 Bilateral debt Multilateral 8000 7000 poverty 6000 goals 5000 sustainability 4000 3000 macroeconomic millennium 2000 1000 0 1995 1996 1997 1998 ...»
Beginning with extreme poverty, it must be noted that the cost of halving the 1990 level by 2015 is defined and gauged in terms of the amount of GDP growth required to deliver the required outcome. Mphuka (2005) shows that halving poverty in accordance with the MDGs target (i.e., bring extreme poverty down from the 58.1 percent level in 1991 to 29.1 percent by 2015) would require a sustained GDP growth rate of 9 percent per year. As Mphuka rightly observes this sort of growth would be unprecedented in Zambia. Although the Zambian authorities do not envisage achieving this growth rate, they are nevertheless optimistic about the country’s growth prospects, expecting that GDP growth will average 7 percent per annum during 2006-2011. More detailed consideration of the financing requirements for reducing the 1991 level of poverty by
17 Zambia: Debt Strategies to Meet the Millennium Development Goals
half is made in Section 2.4 below.
On the sectoral financing requirement of the other MDGs, the costs have been variously estimated as presented in Table 2.1. The costs were obtained largely from Mphuka (2005), with additional cost estimates on HIV/AIDS from NAC (2006) and MOFNP (2006b), as explained in Table 2.1.
The costs in Table 2.1 imply that the overall average investment required in capital 2 and operating expenditure to meet the MDGs is US$1.5 billion per year (over 2005or US$110 per capita per year. Assuming, as Mphuka does, that the MDGs would be largely domestically financed through government contributions of 10 percent of GDP and household contributions of 2 percent of GDP, the total domestic contribution would potentially be 12 percent of GDP. This implies a financing gap of about 13 percent of GDP (or a total of US$803 million annually i.e., avg. of US$56.7 per year per capita).
* The MDGs do not present any specific goal or target with the label “health”; thus the health costs are presented indicatively for insight, not as core MDG cost per se ** Mphuka (2005) does not present child mortality costs separately as they represent bear minimum costs of immunization. Here they are drawn out separately simply to illustrate the minimum indicative costs of child morality as a goal falling squarely into the category of MDGs per se *** These data represent alternative cost estimates as worked out by the NAC (2006) and MOFNP (2006b), respectively. The figures have been re-worked to fit the cost definitions in the table.
Source: Constructed using inputs for Mphuka (2005), NAC (2006) and MOFNP (2006b), with modifications
Source: Author’s construction using data from FNDP and Mphuka (2005) with modifications As a very crude comparison, the main thing to note in Table 2.2 is that the MDG cost estimates as captured by Mphuka (2005) are less than the FNDP cores cost estimates, implying that financing the FNDP core costs would result in a surplus (or excess) of resources over and above the resource requirements for meeting the MDGs. Based on these two separate estimates, the implied resource surplus would be approximately 0.6 percent of FNDP projects of GDP.
However, considerable caution must be exercised in comparing the alternative cost estimates in Table 2.2. This cannot be overemphasized because the estimates have very limited statistical comparability. A number of points validate this assertion: firstly, in Mphuka (2005), the cost estimates do not account for key MDGs cost items such as Goals 7 and 8 on ensuring environmental sustainability and developing a global partnership for development, respectively (see Table 2.1, which misses these). In addition, the costs of other MDGs are underestimated, as demonstrated by the higher alternative costs on HIV/AIDS in Table 2.1 (a detailed assessment is offered in the latter part of Section 2.3 below). It is also noteworthy that costs of technical cooperation for capacity building and other purposes, emergency assistance or other ODA that does not
19 Zambia: Debt Strategies to Meet the Millennium Development Goals
directly finance the capital or operational costs of MDG interventions are not included in Mphuka’s (2005).
Secondly, in relation to the FDNP, various aspects point to cost underestimations.
To begin with, the core FDNP cost estimates do not include the costs of the earlier-mentioned large capital program in the energy, transport and communications sub-sectors;
these are envisaged to be financed through private-public partnerships. This costing 2 approach most likely causes the FNDP estimates to underestimate the true costs. To some extent consistent with Mphuka, the FDNP costing essentially excludes the general administration, running expenses and personnel related expenditures of all implementing sectors except health and education. Perhaps more importantly, the coverage of core FNDP costs is by definition much broader than the cost coverage of the MDGs, particularly as captured by Mphuka. The FNDP costs cover macroeconomic and financial management, central administration, mining, manufacturing, public order and safety, defence, etc. These elements jointly account for a sizable amount (10.3 percent) of core FNDP costs. Including these costs has the bias of overstating the costs compared to the costs of the MDGs.
The foregoing simply highlights the difficulties in getting consistent and comparable cost estimates. To further demonstrate that the financial requirements (or cost) estimates considered above are underestimates of the true MDG costs, Section 2.3 below offers a detailed assessment of the various estimates, also paying attention (where data allow) to the underlying assumptions in each estimation procedure.
2.3 Assessment of the Various Estimates Our assessment of the key issues seen in the financing requirement estimates and their assumptions looks at both the FNDP and MDG cost estimates that this study was able to find. The ensuing discussion is simply meant to draw out a few illustrations that emphasize the magnitude of problems related to financing the development and poverty reduction programs. The assessment focuses on highlighting and commenting on some of the variations in the requirement estimates derived by the observers.
Assessment of FNDP Cost Estimates To begin with, although the FNDP is somewhat modest about the country’s prospects for increasing domestic revenues, the proposed strategies for generating revenue, particularly tax revenue raise questions about feasibility and equity. For instance, looking into the recent history of domestic revenue performance (Figure 2.1) shows that between 1996 and 2005, annual domestic revenue was 18.9 percent of GDP on average. Over the last four years this average outturn has taken a slightly downward trend, averaging 17.3 percent of GDP per year during 2002-2005. FNDP estimates are consistent with this observation showing that domestic revenues currently remains steady at around 18.3 percent of GDP compared, for instance, to the PRSP target of 20 percent by 2004. Tax collections for the period of the PRSP (2002-2005) averaged 17.5 percent and generally
20 Zambia: Debt Strategies to Meet the Millennium Development Goals
stagnated at this level.
According to the FNDP, this revenue performance outturn mainly reflected the difficulty of bringing more segments of the economy into the tax net especially the agricultural sector and the informal sector. It is intended that more pragmatic steps will be taken towards making these sectors taxable. It is expected that ultimately over the FNDP period domestic revenue will increase from 18.0 percent in 2008 to 18.6 percent 2 of GDP by 2010. Achieving this and ensuring the reversal of the marginally downward trend of actual domestic revenue generation exhibited in Figure 2.1 will require significantly better domestic revenue outturns than what the recent trends show.
5 0 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 However, how exactly this could be achieved is unclear. The main proposed strategy for increasing domestic revenue is to include more of the agricultural sector and the informal sector under the tax bracket. This strategy has two key problems: firstly, it might be difficult to execute the strategy given the inherent evasiveness of accurately identifying and being able to actually tax the targeted economic agents in both sectors. The revenue authorities usually have little capacity for tracking the activities of small-scale farmers and subsistence farmers16 and informal sector actors. And even when the capacity for tracking exists, the transactions costs of chasing up tax obligations against meager and uncertain incomes usually outweigh the revenue benefits of tax collection.
Secondly and perhaps more importantly, the appropriateness of a strategy that brings the informal sector and more of the agricultural sector under the tax brackets is questionable. The strategy may have serious poverty and inequality implications, as the average Zambian income tax burden is already disproportionately carried by the poor.
16 Note that in agriculture, it is mainly small-scale and subsistence farmers that cannot be captured in the tax bracket as they are able to easily avoid or evade taxes.
Moreover, the effective tax burden decreases with total income; hence, income taxation is regressive (Gortner, 2004).
Taxes on agriculture and the informal sector would mainly take away from the incomes of the poor as these sectors are precisely where the poor participate, the urban poor dedicating their labour in the informal sector and the rural poor participating in subsistence agriculture.
2 The two illustrative issues brought out above raise questions about the attainability of the domestic revenue projections in the FNDP. This underscores the fact that though the authorities already anticipate financing to be difficult, it is likely the financing will be even harder than the current understanding suggests.
Given these observations, it will be important for the authorities to consider alternative tax policies for increasing domestic revenue. For instance, closer attention should be paid to increasing taxation of the mining sector, particularly now that sufficient time has passed for mining firms to recover some portions of their capital investment. In this regard, there is still considerable debate over the rate of royalty taxes that the government should apply. Many stakeholders, with civil society at the forefront, have expressed that royalties are a viable and sustainable source of national financing, but the government has been slow to respond in terms of formulating an appropriate royalty tax policy.
Reducing the widespread application tax holidays, tax rebates and other tax concessions has the potential to improve domestic revenue generation since mining is such a significant economic activity. Imposing one-off, windfall taxes on mining is another option that could provide the occasional one-off boost to domestic revenue.
Another option might be to pay closer attention to understanding and possibly expanding the list of luxury goods that could attract specific excise taxes. This effort could be complemented with strategies of increasing the taxes on excise duties. However, it should be emphasized that empirical understanding of the costs and benefits of both expanding the list of taxable luxury goods and increasing excise duty rates should first be fostered.
Ultimately, it is up to the finance ministry working with the revenue authorities to identify feasible (i.e., cost effective), progressive, equitable and sustainable tax points within the Zambian economy, which will improve domestic revenue and minimize adverse effects on human development.
Assessment of MDG Cost Estimates In relation to the MDGs cost estimates, the observations made in Mphuka (2005) on the required extent of GDP growth (9 percent per annum) were based on McCulloch et al (2000). McCulloch et al use CSO official poverty data for 1991, 1996 and 1998 to estimate elasticities of extreme poverty to GDP growth and to inequality. Based on the estimated elasticity of poverty to growth, they projected the growth rate and growth path that would be required to reduce poverty from 46 percent in 2003 – the base year
22 Zambia: Debt Strategies to Meet the Millennium Development Goals
in McCulloch et al – to 29 percent in 2015. They determined that this growth path would require sustained growth of 9 percent per year. Two key observations about this finding are noteworthy: firstly, this type of high and sustained growth is unprecedented in Zambia, and is significantly higher than the best-case growth scenario of the FNDP, which estimates that growth under the accelerated FNDP would average 7 percent per annum during the period between 2006 and 2011. The FNDP growth target would 2 itself be unprecedented for Zambia and yet with the weak historic links between growth and poverty, such growth (of 7 percent) would still be insufficient to realize goal number 1 of the MDGs.
Secondly, the required growth rate estimated as 9 percent per annum in McCulloch et al was based on the assumption that the level of extreme poverty at the beginning of the analysis reference period was 46 percent (taking 2003 as the base year). Unfortunately however, the LCMS IV shows that extreme poverty was 53 percent a year later in
2004. This disparity mainly results from differences in the sampling and data collection methodology of the two surveys, the LCMS III of 2003 and the LCMS IV of 2004. In this regard, the CSO has cautioned that in its design the LCMS III was not meant to be and is not comparable with other surveys such as the Priority Surveys (PS I (1991) and PS II (1993)), the earlier LCMSs (LCMS I (1996) and II (1998)) and the more recent LCMS IV. The LCMS III was specifically intended to capture and illustrate intrayear seasonal aspects about the variables and indicators it considers, including poverty.
Perhaps due to the lower level of extreme poverty observed in the survey, many observers gravitated towards anchoring their poverty analyses around the LCMS III. This caused a downward bias in the level of poverty compared with the actual historic levels that are reveled in the other survey datasets. The downward bias is illustrated in Figure 2.2, which compares (in two panels) the poverty trends without and with the inclusion of LCMS III data in 2003.