Confronting South Africa’s Water Challenge: A Decomposition Analysis of Water Intensity

Water resource intensity measures the intensity of water use in terms of volume of water per unit of value added. It is an internationally accepted environmental indicator of the pressure of economic activity on a country’s water resources and therefore a reliable indicator of sustainable economic development. The indicator is particularly useful in the allocation of water resources between sectors of the economy in water stressed countries like South Africa. The study focuses on economy-wide changes in South Africa’s water intensity using both decomposition and regression analysis in an e¤ort to identify and understand the impact of economic activity on changes in the use of the economy’s water resources. The regression analysis suggests that in the long run, water use in the economy of South African is related to the capital stock and that the capital stock is getting more water e¢cient over time, but that this change is gradual. The study results caution against monitoring South Africa’s water/output ratio, and drawing strong conclusions from this regarding changes in water use intensity as the interpretation of this indicator is fraught with di¢culties. JEL Classi...cation: Q25


Introduction
Put bluntly, South Africa does not have enough water. South Africa is ranked as the 30th driest country in the world with a mean annual precipitation of 450mm, just over half the world average of 860mm. This average, however, hides the signi…cant variability in rainfall across the country with less than 100mm falling along South Africa's west coast and more than 1,000mm falling on the east coast per year. The country is also susceptible to periodic and sometimes long-lasting droughts. To aggravate the situation, the mean annual evaporation varies between 800mm and 2,000mm, exceeding the annual rainfall substantially in some areas of the country (DWA, 2013). Indeed, in large parts of South Africa, much of the rain that reaches the ground soon evaporates and re-enters the atmospheric phase of the hydrological cycle. South Africa's ground water resources likewise are scarce, as most of the country is made up of hard rock formations that do not contain major ground aquifers that can be used on a national scale. It is estimated that only 20 percent of South Africa's ground water can currently be used. Ground water resources are used extensively in rural and arid areas and it is estimated that about one-thirds of the population are dependent on ground water for domestic needs (DWA, 2013). With just over 1,100 kiloliters of available freshwater for each person each year, and with a population of approximately 46 million, the country is on the threshold of the internationally used de…nition of water stress. Within a few years, population growth will take South Africa below this level. Indeed, South Africa already has less water per person than countries widely considered being much drier, such as Namibia and Botswana (UNESCO, 2006) For an economy, water certainly is an essential resource in sustaining the living needs of its population and as an input in the production of goods and services in meeting economic growth and development goals. It is vital therefore that water is utilised optimally in order to manage e¤ectively the delicate balance between resource sustainability and that of economic growth. Increasing the output of an economy means more water resources are required as it is highly, positively correlated to economic output. Conversely, the water resources in a region have a natural limit making it impossible to supply without constraints on economic growth. Whilst it is agreed that decreasing water intensity and conservation e¤orts are the most cost e¤ective and simplest of available methods to meet South Africa's future water needs, there is a lack of consensus pertaining to the actual drivers of economy wide water intensity changes and the optimal policies required for the e¤ective management of the country's scarce water resources. Over time two approaches to measuring water use intensity have emerged these include: the engineering approach and the economic approach. The engineering approach focuses on the abstraction, storage, distribution, treatment and disposal activities related to the hydrological cycle and its variability. This approach gives rise to recommendations for the implementation of supply-side water resource management measures, such as infrastructure expansion and investment in reducing leakages. On the other hand, the economic focus on water intensity deals with e¤orts to improve the social gains from water use. It attempts to optimise the bene…ts of the allocation of an exogenously given amount of water in an economy under alternative institutional policies. The focus of the economic approach is on the analysis of water as a consumption good and as an input in production processes and forms the basis of the introduction of water demand-management policies.
This study examines the economy-wide drivers of water use intensity in South Africa over the period 1980 to 2012 from an economic perspective. The analysis highlights the impact of the level of economic activity on changes in water use intensity and attempts to identify possible factors that e¤ect actual levels of water use within the economy. Water use intensity is an internation-ally accepted environmental indicator of the pressure of economic activity on a country's water resources and a reliable indicator of sustainable economic development informing policies of water allocation amongst competing activities in water stressed regions (UN ESCAP, 2009). Information obtained from a decomposition and regression analysis of changes in water intensity should help inform South Africa's water conservation and resource management policies.
The remainder of the paper is set out as follows. Section two provides an overview of South Africa's water usage and the economic theory behind water use intensity. Section three covers a discussion of the decomposition methodology and empirical approach adopted in the study in the examination of changes in South Africa's water use intensity. The results relating to the drivers of changes in the country's water intensity are discussed in section four. The …nal section o¤ers a discussion of policies relevant to the management of South Africa's water resources in light of the study's …ndings.

Background
It is estimated that South Africa's mean annual precipitation runo¤ is approximately 49 billion m 3 per annum of which a mere 10.24 billion m 3 is available annually at high assurance. According to the Department of Water A¤airs (DWA) 9.5 billion m 3 of this available resource is required annually to satisfy the country's total ecological reserve requirement which includes the key ecosystems, namely the rivers, lakes, wetlands and estuaries requiring protection (DWA, 2013). South Africa's water usage typically comprises surface water (77%), groundwater (9%) and re-use of return ‡ows (14%). The water reserve required to satisfy basic human needs (of 25 litres per person per day) translates to 472 million m 3 currently (representing 11 percent of the country's total residential w ater use in 2012). Annual ground water usage in South Africa is estimated to be 2 billion m 3 although it has been suggested that the potential reliable yield is 5 billion m 3 per annum. The annual return ‡ows from irrigation, urban domestic uses and bulk industrial and mining e-uents o¤er re-use opportunities estimated to be close to 1.9 billion m 3 . These numbers put South Africa's total reliable water yield (with 98 percent supply assurance) at close to 15 billion m 3 at present (DWA, 2013). According to the DWA's Water Authorisation and Registration Management System (WARMS) database, total registered water usage in 2012 was estimated between 15 and 16 billion m 3 . This implies that South Africa's current water usage exceeds the reliable annual yield highlighting the urgent need for e¤ective water conservation policies to help address the country's water shortage.
South Africa's water use by economic activity (shown in …gure 1) is dominated by agricultural use for irrigation purposes (63%), livestock watering and nature conservation (3%), mining (2%), industry (9%), commercial and business use (7%) and residential use (16%). The percentages indicated are for the year 2012 (DWA, 2013) In 1998, the South African government promulgated the National Water Act (DWAF, 1998), which recognised water as a national asset and a strategic resource for the country's economic and social development. The National Water Resource Strategy (DWAF, 2004a) and the Waste Discharge Charge System (WDCS) (DWAF, 2000) were developed to help implement the National Water Act in addressing the management of the country's water resources to meet her development goals. Historically the management of South Africa's water resources has focused on the supply-side. More recently however, the DWAF (2004a: 78) has recognised that, "the options for further augmentation of water supply by the development of physical infrastructure are limited and in the future attention will have to be on managing the increasing demand for water in order to achieve a sustainable long-term balance between water availability and water requirements." The options for the supply-side management of South Africa's water resources include: inter-basin water transfers, the desalination of seawater and acid mine drainage. These supply-side engineering solutions "are however becoming less viable, and water managers are turning to the attractive solutions o¤ered by demand-side management" (King, 2004: 208).
The key to the management of South Africa's scarce water resources therefore lies with demand-side approaches. The South African government recognises this and explicitly states in its National Water Resource Strategy that "water-demand management and water conservation can be achieved through the e¢cient use of water combined with pollution abatement, re-use and recycling of water and water-e¢cient technologies" (DWAF, 2004b: 6). Water use intensity has its origins in the economic concept of productivity. In economics productivity measures the amount of any given resource that must be expended to produce one unit of a good or particular service. Water use intensity is the inverse of water productivity which measures the value added generated by one unit of water used. Water productivity gives an indication of the intrinsic value being placed on water. It has low values when water is used for low value purposes, which is generally the case when water is abundant and/or undervalued. High values of the indicators are associated with water recycling and improved technology which reduce the amount of water used and therefore abstracted (UN, 2006). Water use intensity per economic activity in South Africa should thus in a similar manner be measured by the volume of water utilised to produce a unit of output, were the output is commonly measured in value added terms (see Table 1). In general, the lower the water input requirement per unit of output, the lower the water intensity associated with an economic activity.
In an environmental economic context, the resource intensity concept must however be extended to include considerations of quality (UN ESCAP, 2009). Any e¤ort to manage the country's water resources to reduce water use intensity should also take into consideration e¤orts to maintain or improve water quality in South Africa. Information on water use intensity is important in informing water conservation e¤orts. The latter term has received many de…nitions in the past, the de…nition used by Baumann et al. (1980) is most relevant to the present study, namely, water conservation involves a socially bene…cial reduction in water use or water loss. In this context information on water use intensity is of central importance in highlighting economic pressures on water resources.
Similarly, the conservation de…nition suggests that the intensity measure should, in addition to reducing water use per unit of activity, make sense economically and socially within an economy (Tate, 2000) In summary, the reduction of water use intensity in South Africa should involve increasing water productivity through reducing the intensity of water use, improving the allocation of water among competing water uses so as to obtain a greater socio-economic value per drop of water utilised and improving the technical e¢ciency of water service provision within the country.
Economic factors are amongst the most important drivers of changes in water use intensity Water has two main uses in an economy; it is utilised directly as a consumption good by households or it is employed as a factor of production in agriculture, forestry, mining, industry and commerce. The theoretical foundations of these uses di¤er and are discussed separately. Residential demand is the only instance in which water is consumed directly. Residential water competes directly with other items in the household budget of South Africans. Consumer choice can be modeled as utility maximisation given a budget constraint from which a downward sloping demand for water can be derived. A condition for economic e¢ciency in consumption is that marginal utility must be equated for all consumers. This principle is generally achieved in South Africa as all consumers in a given area face the same price for water. Turning to water as an input in production, theoretically its demand is dependent on the demand for the product it is used to produce, the water intensity of the production process and the available water-saving technologies. Water's use in this context should be related to the ratio of its marginal productivity per rand of value added in production relative to that of other inputs, namely, capital, labour, energy and raw materials Water has traditionally been regarded as a public good, which makes management strategies of water pricing and allocation an extremely sensitive topic. These strategies are not normally based on economic e¢ciency but rather on social and distributional criteria such as fairness and/ or equity considerations (Moolman et al, 2006). A variety of mechanisms are used in countries to allocate water resources among competing uses. These range from complete government allocation, to a mixture of market and government allocation, to predominantly market allocation. Dinar et al. (1997) distinguish between four di¤erent water allocation mechanisms which include: (1) public (administrative) water allocation, whereby the state decides how much water is allocated to di¤erent uses; (2) user-based allocation schemes, extensively used in agricultural irrigation systems; (3) water markets, where trading occurs in water rights; and (4) pricing, where water use is charged. With increasing water scarcity and acknowledgement world wide that water has an economic value in all its competing uses and as such should be recognised as an economic good (UN, 2006), authorities within countries are increasingly turning to market allocation approaches which include assigning water rights and pricing.
Market prices will result in an e¢cient and equitable allocation of a good, service or resource if it is correctly priced. Under conditions of economic scarcity, value is attached to water resources and access rights are assigned to water 5 resources. When these rights are traded, a price is negotiated in the market transaction. In economic theory, the optimal combination of inputs, or "economic e¢ciency", occurs when the marginal prices of each of the factor inputs are equal. If any required input has a very low, or zero price, to the user, as much of that input will be used as required. Water belongs to a group of materials called common property resources, to which access is nonexclusive, ownership is held in common by the public and prices are very low, or zero. Stated simply, when water prices are low relative to the costs of other inputs and in relation to the costs of developing supplies, water resources will be overused and water intensity will be correspondingly high. Regardless of their origin, low water prices are the enemy of e¤orts to conserve water use in that there is less incentive for users to recover, recycle and decrease their demand for water, as compared to a situation in which they have to pay more for water (Tate, 2000) Water markets are however not suited as pricing strategies for all uses. In the case of household use water markets will result in high value users bidding up the market clearing prices that will have social costs on low-income consumers in that they will be forced to spend large proportions of their income on water (Moolman et al, 2006).
Generally water prices are de…ned as the …nal price that the end-user pays per meter cube (m 3 ) of water. It is acknowledged however that there is no such thing as a single price for water in South Africa and that the end-user price is constructed on the basis of a variety of components, and water tari¤s. Currently end-user prices for water in the country include a "water resource management" charge, a "water resource development and use of water works" charge, a "water research fund" levy, as well as bulk and/or retail water tari¤s (Eberhard, 2003). By and large, these charges and tari¤s are based on cost recovery principles. The …rst two of these charges varies according to geographical area and economic sector and is designed to recover the costs associated with water resource management in the …rst instance and the construction, operation and maintenance of water supply schemes and infrastructure in the later case. The third charge is a levy earmarked to fund the operations of the country's water research commission. Bulk tari¤s are designed to recover the costs associated with raw water abstraction, bulk water treatment and distribution, while retail water tari¤s are designed to recover the costs of reticulation of water to consumers (Eberhard, 2003). Historically the costs of providing water services in South Africa have not been fully recovered. It has been noted in many studies that that the …nal price paid by water users in South Africa does not re ‡ect the value derived from water, the opportunity costs of water use, or the scarcity of water resources (Cummings and Nercissiantz, 1992;Dinar and Subramanian, 1998;Eberhard, 2003). An excessive quantity of water is utilised in the economy leaving little to sustain ecosystems and at the same time large quantities of wastewater and pollution are discharged into the country's surface-and ground-water sources.
6 Figure 2 shows changes in South Africa's aggregate water use intensity for the period 1980 to 2012 (the period for which reliable water use data per broad economic activity could be sourced from the records of the Department of Water A¤airs formerly know as the Department of Water A¤airs and Forestry). The trend in the country's water use intensity appears to be decreasing through the years. This should not however be interpreted as a reduction in water consumption over the study period. Both value added and water consumption increased through the years but the increase in water consumption was lower than the increase in economic output so the overall water to output ratio decreased.
The water intensity value shows how many units of water (measured in m 3 ) are utilised in the production of economic output (measured in ZAR millions). The total economy decreased its water utilisation in the production of R1 million of ouput from 7,160 m

Decomposition through the Fisher Ideal Index
The study adopts the Fisher Ideal Index approach to perform the decomposition task at hand. The main advantage of this method is that it does not involve any residual terms, which would make it di¢cult to interpret the relative importance of structural and intensity (technical) e¤ects. Speci…cally, Ang and Zhang (2000) emphasise that perfect decomposition methods are preferred in the case of two-factor decomposition due to their theoretical foundation and their adaptability, as well as the ease in interpreting their results. In this study, changes in South Africa's water use (w t ) are decomposed into structural and intensity components. Following the decomposition literature, the research problem is set in terms of total water use (W ) and total production (Y ), as well as sub-indexes for economic sector (i) and years (t). Thus, the aggregate water use intensity (w) can be written as: Equation 1 indicates that a change in w may be due to changes in water intensity (w it ) and/or the structural composition (s it ). One of the main prac-tical advantages of this approach is that, by construction, the water uses in the di¤erent sectors need to form a partition (i.e., they must not overlap), but the measures of economic activities do not need to satisfy this condition. In fact, they do not even need to be of the same units. This facilitates the identi…cation of good indicators to account for the structural composition (s it ). Following the theory on index numbers, dividing equation (1) by the aggregate water intensity for a base year (w) allows a perfect decomposition of the aggregate water use intensity index into intensity (F int ) and structural (F srt ) indices with no residual. The result, the Fisher Ideal Index, is a geometric mean of the Laspeyres and Paasche price indices.
The Laspeyres indices are: and the Paasche indices are: The Laspeyers indices use a base period …xed weight while the Paasche indices use an end period. The Fisher Ideal Indices are then given by: Fischer (1921) showed that this index satis…ed perfect decomposition of an expenditure index into a price and quantity component. In the context of this study, the Fischer Ideal index provides a perfect decomposition of South Africa's aggregate water use intensity index.
By taking the logarithm of equation (4), it is possible to observe the additive contribution of the structural e¤ect (F srt t ) and the intensity e¤ect (F int t ) to the total variation in water use per output unit (I t ). This decomposition suggests a way to attribute changes in water consumption arising from improvements in water use intensity. Water savings (¢Wt) due to changes in water intensity are then de…ned as: where W t is actual water use and Wt is the water use that would have occurred had water use remained at its 1980 level. The study attributes the change in water use between intensity and economic structure as follows:

Empirical results
First up are the results of the Fisher decomposition analysis of South Africa's water use intensity computed by partitioning the country's aggregate water use into agricultural, mining, industrial, commercial and residential sub-sectors through the assignment of appropriate economic activity measures to each of these water using sectors (refer to the Appendix 1 for further detail). Thereafter the results of the decomposition analysis are interrogated through regressions techniques to determine the main drivers of changes in South Africa's aggregate water use intensity to be 44 percent of its intensity level in 1980. Using equation (6), the study allocates the change in the country's water use (relative to the amount that would have been consumed had water use remained at its 1980 level) between intensity (technical) and structural e¤ects. The structural index is 93 percent of the level it was in 1980 while the intensity index is 51 percent of its 1980 level. In other words, had the composition of economic activity not changed between 1980 and 2012, water use intensity would have been 51 percent of its 1980 level. The 49 percent reduction in South Africa's aggregate water use was due to decreases in water intensity. Similarly, had water use intensity been …xed at its 1980 levels for all sectors, changes in the structure of economic activity would have led to a 7 percent decrease in water use Based on the decomposition analysis, the bulk of the 66 percent reduction in water use arises from changes that can be attributed to decreases in intensity of use. In contrast changes in the composition of economic activity in South Africa have only a minor impact on the country's aggregate water use Figure 4 shows the contributions of changes in intensity of use and the structure of economic activity on water savings between 1980 and 2012. Through out the period under investigation changes in water use can be attributed almost entirely to reductions in intensity of water use. It is important to highlight that the decomposition is conditional on the particular choice of water using sectors identi…ed in the analysis used in the partitioning of the data.

Drivers of South Africa's water use intensity
The decomposition results highlight the signi…cance of technological improvements in driving changes in South Africa's water use intensity. The intensity (or technical e¤ect) is the dominant factor that contributes to downward pressure on the country's water consumption. This is because the intensity e¤ect works in either of two ways (or a combination thereof), namely, (1) technical progress can motivate users of water to substitute other production inputs for water resources and/or (2) it could encourage them to decrease their water usage through the recycling of water use. Policy makers should, therefore, implement appropriate policies to promote technical progress and the re-use of water resources Whilst reductions in water use intensity embodied in new capital may well account for a proportion of the intensity (technical) e¤ect, one would still expect this e¤ect to be relatively small and constant over time, and not a driver of the large short run ‡uctuations in the intensity e¤ect.
Turning the study focus on the intensity e¤ect, the decomposition analysis seems to show that water intensity rises during periods of low economic growth and falls in periods of high economic growth. It would seem from casual observation that this e¤ect is due to higher levels of investment in periods of high economic growth. With the new capital stock being less water intensive than that which it replaces, there is then a corresponding marked improvement in water use intensity in the economy. Such an explanation does not however appear to make intuitive sense. This is because the proportion of the total capital stock replaced in any year, even in a year when economic growth is high, is relatively low. Furthermore, it is not clear why water use intensity would rise even in periods when investment is low. For even in years when investment in South Africa has been low, the level of investment has been more than enough to o¤set depreciation, therefore lifting the country's capital stock.
On closer examination it appears that variations in South Africa's water use intensity are more likely due to changes in capacity utilisation (see …gure 5). A relatively high proportion of water use in economic activities is …xed, being used for cooling, heating, cleaning purposes etc. The remainder re ‡ects changes in production levels, and represents the variable proportion of total water costs. It is changes in these variable costs that re ‡ect short run changes in the aggregate level of economic activity. Because of the …xed component, total water use will not ‡uctuate as much as output. Hence water use intensity is relatively volatile, re ‡ecting changes in output or capacity utilisation. The study derives capacity utilisation for the economy by taking the ratio of capital stock to output. In fact the ratio represents the inverse of capacity utilisation. The household sector is excluded from the analysis, by using gross …xed capital formation excluding residential investment within the economy. Figure 5 indicates that capacity utilisation and the intensity e¤ect track each other fairly closely. Recall, that the intensity (technical) e¤ect index represents the movement in the residual component of the water to output ratio after adjustment for industrial structure. Hence, the intensity (technical) e¤ect index is a good indicator of the intensity of water use. Note that a rise in the capital/output ratio indicates a fall in capacity utilisation, and vice versa. Hence, Figure 5 reveals that the water/output ratio falls as capacity utilisation rises, and rises as capacity utilisation falls. This trend is consistent with the notion of having separate …xed and variable costs associated with water usage.
What still remains unexplained is why the water/output ratio would actually rise during periods when net investment is positive. To interrogate this the study runs a number of regressions, to verify the magnitude of the various relationships. The regression variables and equations are set out in Appendix B. Regressing the intensity e¤ect on the capital/output ratio generates a R 2 of 0.74. Regressing the intensity e¤ect on investment gives a R 2 of only 0.07. The R 2 of both series together against the intensity e¤ect is 0.82. To check that this is not a result of multi-collinearity between investment and the capital/output ratio, the one is regressed against the other; at 0.004, the R 2 does not suggest a signi…cant relationship between these two series. The regression analysis therefore indicates that the intensity e¤ect, or water use per unit of output, is a¤ected by both capacity utilisation and investment, with the equation being: where: IE is the intensity (technical) e¤ect index KS=Y is the real productive capital stock/output ratio GF KF is the real gross …xed capital stock (investment). Given that the equation's coe¢cients are in logs, we can interpret these as elasticities. Hence a 1 percent rise in the capital/output ratio can be expected to result in a 1.1 percent increase in the water/output ratio, while a 1 percent rise in GFKF will result in a 0.24 percent fall in the water/output ratio. Hence both factors are at work. A rise in capacity utilisation will lower the intensity of water use, as will a fall in investment. The analysis thus suggests that when investment levels in South Africa are lower than in the previous year, the contribution from investment causes a rise in the water/output ratio. This development still appears to be counter-intuitive. As mentioned earlier, even in years when investment falls in South Africa, the level of investment is still substantial. It therefore does not seem to follow that an easing in the level of investment will necessarily result in a rise in the water/output ratio. On closer inspection investment is found to be highly correlated with output, with an R 2 of 0.85. Hence, investment may have entered into equation (7) because of its correlation with output, which is the denominator of the dependent variable in this equation, namely: the water to output ratio. In an initial attempt to remove this e¤ect, the regression analysis included dividing GFKF by output (see equation 5 in Appendix B). This did not however have a notable impact on the performance of the regression equation.
Suppose that the output level is not used as the denominator of water use (the dependent variable) in equation (7) but instead the regression analysis simply represents the variables of interest in the study in levels. That is, the analysis represents water use as a function of the capital stock and the investment level, as in equation (8).
This representation of water use makes intuitive sense, especially if the economy's capital stock is lagged by one year. That is water use is then dependent on last year's capital stock and this year's increment to the capital stock (investment). This suggests that the capital stock level generally determines water use. But does the data back this up? Figure 6 seems to support this view. In contrast, the correlation between water use and output is not as high (see Figure 7). Water use and output tend to diverge in the early 1990s when capacity utilisation was low in South Africa.
In fact, this chart sheds light on changes in the intensity (technical) e¤ect (the water/output ratio) as reported earlier. It seems clear now that the rise in the water to output ratio in the mid 1980s and early 1990s was due almost entirely to falls in output. (The regression results for the intensity (technical) e¤ect equation picks up this relationship via capacity utilisation, which was low in this period.) If as the regressions analysis suggests, namely that water use generally re ‡ects the level of capital stock, this would support the view that the …xed component of water use is relatively high. In other words water use in economic activities is related to the machinery, equipment and infrastructure that is in operation, rather than being simply a function of the output that is being generated. In view of this, the study adopts a co-integration approach, estimating a long run equation relating water use to the capital stock, output, and capacity utilisation. The best …t in respect of the long run equation relates the level of South Africa's water use to the economy's capital stock (see Appendix B). The R 2 for this equation is high, at 0.92. This supports the view expressed above that water use is generally related to the level of capital stock. An interesting feature of the long run equation is the coe¢cient on the capital stock, which at 0.963, is less than 1. This implies that a 1 percent rise in the capital stock will be accompanied by a 0.96 percent rise in water use. Hence, the equation shows that the capital stock is getting less water intensive although this change is slow.

Conclusion
In the face of South Africa's growing water scarcity, water conservation and the e¤ective management of water use should be priority focus areas of water policy in the country. There exists an urgent need to introduce market based incentives to optimise the allocation of scarce water resources between competing uses. These economic incentives should target water-stressed areas with the objective of encouraging a shift of water use from economic activities with low water e¢ciency values to activities with high water e¢ciency values.
Unfortunately, under South Africa's previous water legislation, the pricing of water did not take into account either the real cost of managing water, the cost of water supply or the scarcity value of water. MacKay (2003) suggests that the capital costs of government water schemes supplying mainly agricultural water users (and many urban bulk water suppliers and industrial users) were either …nanced or heavily subsidised by the government. Additionally, operation and maintenance costs were often not fully recovered from these water users (MacKay 2003:64).
Ultimately the management of South Africa's water resources should involve using information obtained from an analysis of water use intensity to design and develop policies to help allocate water to its most productive use, whilst maintaining ecosystem functions, human well-being and social equity. The di¢culty lies in the implementation of policies that encourage users of water within the country to use water in a manner re ‡ective of its scarcity value. Such policies will help South Africa achieve the required water savings with minimal disruption to the economy. The set of policy solutions includes both technical improvements to increase supply as well as measures to reduce the intensity of water use.
The regression analysis undertaken as part of the study on economy wide water use intensity has shown that in the long run, water use in the economy of South African is related to the capital stock. The analysis also suggests that the capital stock is getting more water e¢cient over time, but that this change is gradual. The study thus suggests that in the longer run it is the relationship between water and the economy's capital stock that has to be changed, if South Africa is to become more water wise. The regression analysis furthermore suggests that monitoring South Africa's water/output ratio, and drawing strong conclusions from this regarding changes in water use intensity, is fraught with di¢culties.
The study recognises that markets and water prices are not necessarily the ideal mechanism for distributing an essential good such as water. Water authorities in South Africa may have sound reasons for using other allocation mechanisms which include the need to ensure that ecosystems have adequate water to function and deliver services. Only after adequate water resources are made available to meet basic human needs and to safeguard ecosystem health should water be allocated for other uses within the economy based upon the concept of water productivity (UNEP, 2012). It is recognised though that regulatory mechanisms including: monitoring, volumetric metering and enforcement of legal limits on water use and pollution are often costly and varies in its e¢cacy due to inaccurate information between regulators and water users (UNEP, 2012). The study supports the use of a combination of approaches to water resource management, including information-based tools such as labeling and the education of producers and consumers. Operating together, the mixture of water resource management tools should incentivise innovation and the adoption of new processes and technologies, reducing the intensity of water use across all sectors of the economy, thereby assisting policy makers to decouple water use and its impact on South Africa's economic growth.

Appendix A
The study employs in its analysis of South African water intensity, measures of economic activity related to the underlying water use within each of the economy's sub-sector. Based on the methodology employed in Metcalf (2008) personal consumption expenditure is employed as the activity measure in the case of South Africa's residential water consumption. This according to Metcalf (2008) is preferable to disposable income since a portion of disposable income goes to savings that should have no discernable impact on residential water demand. In the case of South Africa's mining, commercial and industrial water use, the study employs the concept of value added for these sectors of the economy. Value added is a measure of the contribution to final production from a given sector. Table A1 identifies the economic sub-sectors for which water use is analysed and presents summary statistics for the measures of economic activity of these sectors. The identified economic sub-sectors are employed for the purposes of the decomposition analysis as well as in the generation of statistics on sector level water efficiency. The table thus also provides summary statistics on the data used to construct the water indices for South Africa. Source: Water use data from DWAF and economic activity data is from Quantec.
(1) Data from 1980 to 2012. Agriculture includes forestry and fishing related activities. The commercial sector includes communication, wholesale and retail trade, finance, other commercial services, and public services. Table A2 presents the water use indices for South Africa over the analysis period.