WATER SUPPLY INTERRUPTION: DOES LENGTH MATTER? AN EMPIRICAL APPLICATION TO RESIDENTIAL USERS IN SEVILLE

Transcription

1 WATER SUPPLY INTERRUPTION: DOES LENGTH MATTER? AN EMPIRICAL APPLICATION TO RESIDENTIAL USERS IN SEVILLE Maria A. García-Valiñas a,b David Roibás a Abstract: EU Water Framework Directive leads to manage water in an efficient way. Water scarcity originates frequently the need for rationing, and various methods, including suly interrutions, can be used. The aim of this aer is to evaluate the effectiveness and efficiency of those methods in reducing residential water consumtion. We estimate a demand function segmented into two comonents in order to cature the roortion of water consumed by households under different suly cut schemes. We find that the reduction in consumtion er hour of cut decreases with the length of the daily interrution. Thus, it is better to imlement many short cuts than a few long cuts in order to achieve the targeted reduction in consumtion, minimizing the total time of interrution. Additionally, we show the relative effectiveness of rices to control water demand. Keywords: consumer behaviour, water services, suly interrution, welfare economics, JEL classification: D11, D60, Q25 a Deartment of Economics, University of Oviedo b Corresonding author: Maria A. García-Valiñas Deartment of Economics, Faculty of Economics, University of Oviedo, Avenida del Cristo, s/n, 33071, Oviedo (Asturias), Sain. Tel: Fax: address: mariangv@uniovi.es

2 1.- Introduction Traditionally, water has been considered as one of the most imortant natural resources and its availability has conditioned the location of economic activity (Gibbons 1986). At the same time, water is a limited resource that has given rise to numerous conflicts over its allocation between alternative uses (Lee 1999). The imortance of water lies in the fact that it satisfies a broad grou of needs, both in its role as a necessary good uon which ublic health and life itself deend, as well its role as a basic inut in most agricultural and industrial roduction rocesses. This double role of consumtion good and factor of roduction makes water an essential factor in economic activity and obliges the authorities to invest in infrastructure which guarantees its availability, often at a high cost. 1 Demograhic ressure and the rocess of economic develoment lead to a continual increase in the consumtion demand for water, while at the same time the climatological conditions which limit the availability of water in many geograhical zones aear to be worsening as a consequence of the rocess of climatic change which the lanet is undergoing. 1 In Sain, for examle, the contrast between regions in terms of the natural availability of water has led to a olicy of diverting water between basins. The National Water Plan aroved in 2001, which aims to imrove the water suly in regions in the south of Sain on the Mediterranean coast, has an estimated cost of around 3.78 billion euro. This Plan was abandoned by the new government in 2004, which suorts conservation olicies and desalinization lants in Southeastern Sain as an alternative. 1

3 Those factors have the effect that eriods of water shortage can be exected to reaear in the future. This in turn will give rise to the need for olicies which limit consumtion. In articular, one of the most serious causes of water shortage in many regions is drought, and when this cyclical henomenon reoccurs the entities resonsible for water suly often imose water cuts in order to match the available suly with demand. Generally, water shortages give rise to the need for rationing, and while the authorities frequently resort to suly cuts, there are various other ways in which this can be achieved (Winenny 1994). EU Water Framework Directive leads to manage water in an efficient way. So it would be really interesting to find out what rationing method is more efficient and effective to control water consumtion. Several studies have focused on the effects of olicies aimed at limiting water consumtion (Moncur 1987; Woo 1994; Renwick and Archibald 1998; García-Valiñas 2006; Roibás et al. 2007). In this aer, we evaluate the imact in consumtion of several rationing regimes: suly interrutions and rice rises. Additionally, we are interested in finding out what kind of hourly interrution framework is more efficient in order to control residential water consumtion. We aly the model to analyze the effects caused by the drought which affected the city of Seville (Sain) in the first half of the 1990s. We roceed as follows. In Section 2 we describe the effects of the drought in Seville in the early 1990s and the main initiatives imlemented by the sulier in resonse. Section 3 resents the theoretical model used for the emirical analysis. Section 4 shows the emirical secification. Data set is described in Section 5, and the estimates and welfare results are shown in Section 6. Section 7 concludes. 2

4 2.- The drought in Seville in the early 1990s Seville is a city in the south of Sain with more than 700,000 inhabitants (INE 2006) which has been severely affected by scarcity roblems several times. The drought during the first half of the 1990s will be the focus of our attention. The shortages aeared at the end of 1991 and the situation worsened the following year, which coincided with Exo The first restrictions on consumtion, formally introduced through the ublication of municial edicts 3, were ut into lace at the beginning of These edicts secified the conditions under which water would be sulied as well as the resonse which was exected from users. The restrictive measures ut into ractice comrised cuts in suly and reductions in water ressure. Hour restrictions were articularly intense during the years 1993 and During this eriod, water cuts sometimes lasted for u to twelve hours er day 4. Regarding the structure of interrutions, the shortest (basic) cuts went from 2 a.m. to 6 a.m. If ublic authorities decided to increase the duration of the daily interrution eriod, then they extended the basic cut towards the last hours of the night and evening. 2 The World Fair which took lace in the city in Although suly cuts are legally ermitted under certain circumstances, it is obligatory to notify users in advance about when the cut will take lace, as well as the intensity and the estimated duration (Molina 2001). 4 Those restrictions were not alied to users, such as health centers, roviding services deemed to be essential to the ublic interest 3

5 Aart from restrictions on the hours of water availability, the quality of the resource was also affected. Desite the efforts of the sulier to guarantee water quality, the secial circumstances of the weather and the consequent decision to diversify the sources of rovision led to a considerable deterioration of the resource. The health authorities found themselves with no other otion but to imlement a rovisional relaxation of water standards using the argument that this was a eriod of excetional conditions. 3.- The model In this section, we design a model which catures the influence of temorary unavailability of the resource on consumer demand. These temorary interrutions have two main observed effects on consumer behavior. Firstly, consumers react to an interrution by acquiring only a roortion of the water that they would have consumed had there been no cut 5. Secondly, the longer the duration of the interrution, the lower the roortion of water acquired. It should be noted that consumers can install devices such as storage tanks to soften the imact of interrutions. These exenditures will in turn generally deend on household income. We therefore exect that the reduction in consumtion due to the suly interrutions may also deend on household income. To cature the aforementioned characteristics, we model a marshallian water demand function comrising two comonents as follows: 5 Consumers are able to maintain their overall water consumtion at re-interrution levels if they so wish (by consuming at times when water is available). Changes in consumtion due to the interrution are therefore a consequence of changes in consumer behaviour induced by the cut. 4

6 [,y,s;f(c, y) ] = q(,y,s) f(c y) Q = Q, (1) The first comonent, q( ), reresents a standard marshallian demand function which deends on the rice of the rationed good (), income (y), and other variables which affect the utility of the consumer (s). The second comonent, f( ), is a function which catures how the duration of the interrution (c) and the household income (y) determine the roortion of the desired good that the consumer acquires when a suly cut takes lace. It is assumed in this model that the amount of consumtion loss due to the suly cut deends on the lanned consumtion and therefore on the variables which determine this lanned consumtion. However, it is also assumed that the consumtion lost as a roortion of lanned consumtion deends only on the duration of the cut and household income. In this sense, the effect of the interrution is similar to that of a roortional rationing scheme (see, for examle, Tirole 1988), under which consumers only acquire a certain roortion of their desired quantity of the good. Figure 1 shows the water demand function consistent with the assumtions maintained in the model under two different scenarios. When there are no interrutions, the water demand curve is reresented by Q, whereas Q reresents the demand function when suly interrutions are in lace. Thus, the introduction of a water cut induces a 5

7 reaction on the art of consumers such that the water demand curve moves from Q toq, reducing the quantity demanded, given the rice, from Q NR to Q R. 6 Figure 1 Welfare losses: good water quality P a v c g d Q = Q[, y, s; f ( c = 0, y) ] [, y, s; f ( c 0, y) ] Q = Q > Q R Q NR Q 6 It is imlicitly assumed in the model that the intercet of the demand function does not change when a suly cut takes lace. This may be questionable in the sense that the intercet of the demand function refers to the most valuable unit of the good and it could therefore be argued that this value may change when an interrution of suly occurs. However, as households are able to store a certain amount of water when the suly cut is announced, this storage caacity will assure the consumtion of the most valuable unit and as a result it can be argued that its value does not change. 6

8 Additionally, under water quality reductions, the rice which consumers will be willing to ay for the first m 3 would be reduced. It is ossible to observe the revious idea in the Figure 2, in which demand function moves taking into account both shortages and quality reductions. Figure 2 Welfare losses: bad water quality P a a v LQ v c f g Q LQ R Q d e LQ Q = Q Q LQ NR LQ LQ Q = Q[, y, s; f ( c = 0, y) ] LQ = Q [, y, s; f ( c = 0, y) ] [, y, s; f ( c > 0, y) ] Q Based on this modeling of the marshallian demand function, we roose using the consumer surlus as an aroximation to the welfare losses associated with the distinct rationing systems. In the basic scenario (good water quality, see Figure 1), the introduction of a water cut has the effect that consumer surlus is reduced by: 7

9 a a [,y,s;f(c =, y) ] d Q[,y,s;f(c 0 y) ]d C W = Q 0 >, (2) We comare the effects on welfare of the interrution with those which would arise from a olicy of rationing via rices. To do so, we need to determine how much the rice should be raised in order to obtain a similar effect to that of the interrution. That is, from the demand function we need to determine the virtual rice (v) which would give rise to a reduction in consumtion equivalent to that which would occur under a given interrution. Raising the rice to v therefore reresents an alternative, but equally effective, method of rationing, and the corresonding loss in consumer surlus would be (area vcd): v P E = Q[,y,s;f(c = 0, y) ]d (3) However, art of this variation is accounted for by a transfer of rent to the sulier. As water rovision is tyically regulated by the ublic sector, these rents may return to consumers through ublic sending. Under this scenario, the net loss in social welfare is: R [ = y) ] d (v ) Q v P W = Q,y,s;f(c, 0 (4) Comaring (2) and (4) using Figure 1, it can be seen that the welfare loss under rice rationing is always lower than that corresonding to an interrution. This result is due to the fact that under a rice rise consumers only acquire units of the good which they value more than v. This imlies that consumers forego the marginal units of water. Under an interrution, on the other hand, they acquire units which they value more than as long as they are available. However, an interrution may revent consumtion of certain units of the good valued by the consumer above v which are 8

10 substituted by other units valued between and v which are available. This imlies that the consumer foregoes units of water which are not necessary the marginal ones, imlying a further loss in consumer welfare. Finally, we adjust welfare losses considering water quality. We are assuming that water sulier is not able to control inut water quality, due to the extreme conditions to suly under drought eriod. Based on Figure 2, welfare losses liked to water quality (area aa ed) can be calculated by the following exression: a a LQ [,y,s;f(c =, y) ] d Q [,y,s;f(c = 0 y) ]d LQ W = Q 0, (5) Additionally, it would be ossible to recalculate virtual rices in the new context (v LQ ). This virtual rice would aroximate the valuation that users give to quality water resource and service, without any quantity or quality restrictions. So, by means of this virtual rice, scarcity costs of water resources could be aroximated. 4.- Emirical secification In order to estimate the marshallian demand in equation (1), we roose the following emirical secification: Q it [ β 0 + β it 2 + β y y i + β n n i + β qual qual t + β tem tem t ] s [ 1 + d ( α + α y + α c + α c )] l = h s s i s t l t (6) Quantity of water (Q) total household consumtion er eriod. The demand function in equation (1) measures is secified as the roduct of two comonents. The first comonent reresents the marshallian water demand in the absence of suly cuts, 9

11 modeled as a linear relationshi, where ( it-2 ) is the rice that the i th consumer ays for water in eriod t-2; (y i ) is the consumer s income, (tem t ) is the average temerature in the eriod t, (n i ) is the number of ersons in the consumer s household and (qual t ) is a dummy variable indicating the resence of water quality reductions during the eriod. The second comonent catures the effect of the interrution on the quantity of water acquired, where (d h ) is a dummy variable which equals 1 when interrutions takes lace. Thus, we have differentiated between a short cut (c s t), which includes the total suly cut duration in the eriod t when marginal suly interrutions are minor or equal than six hours er day, and a long cut (c l t), which considers the global duration of the interrution when daily cuts are longer than six hours er day. This second comonent includes income too. Finally, α and β are arameters to be estimated. 5.- Data set To estimate equation (6), we have used a balanced anel consisting of quarterly observations corresonding to 208 Sevillian households with individual meters, covering the eriod from the fourth quarter of 1991 to the third quarter of The data come from several sources. The Sevillian municial water suly firm (EMASESA) rovides information about water consumtion er household (Q), the rice of water and the resence of restrictions into the quality of the water sulied (qual). The rice is calculated as the average quantity aid er cubic metre, which is regarded as the most aroriate indicator in a framework of imerfect information (Charney and Woodard 1984; Oaluch 1984). The two-eriod lag in the rice is introduced to cature the fact that users only observe the rice when they receive the bill. The comlexity of the water rate structure means that the bill thus becomes a 10

12 crucial informative tool for users. The bill is received, however, during the quarter following consumtion, so consumers can only react to variations in the rice after two eriods, hence the two-eriod lag. Quality variable (qual) is equal to 1 when there are dros in ressure and/or chemical arameters which determine the quality of the roduct. A roxy for the family income (y) is obtained from an estimation erformed by La Caixa for the year 1998 (La Caixa 2000). We use a roxy based on the location of the home within the municiality. The Revenue Office of the Municiality of Seville rovides information on the fiscal category of each street, and for each category an interval of disosable income has been assigned. The estimates for available income are rovided by a research study carried out by La Caixa for the year 1998 (La Caixa 2000). There are eight income intervals in total, and the mean of each interval is used to aroximate family disosable income. The data about temerature (tem) are rovided by the Sanish National Meteorological Institute. We use the arithmetic mean of the maximum daily temeratures during the eriod under consideration. Finally, the number of users er household (n) has been obtained by dividing the total number of eole in each building who aear in the census by the number of households in each building. However, we could get the real number of eole for household in the case of large families (4 o more members), due to the alication of discounts in water tariffs deending of household size. Anyway, difficulties in getting information revent us from determining how the number of eole in each household 11

13 varies over time. Hence, the variable only exhibits variation across the different households and does not change over time. The information on this variable is rovided by Seville City Council and EMASESA. Some descritive statistics relating to the variables and the quantities consumed are resented in Table I. Table I Variables used in the estimations: descritive statistics Variable Units Mean Stan. Dev. Max. Min. Q m Euros/m y Euros/household tem Celsius Degrees n Persons/House c day Hours: Minutes 4:50 2:03 7:00 0:40 Source: own elaboration a When calculating the statistics referring to the duration of the interrution we exclude eriods for which there are no interrutions. 12

14 6.- Results The estimation was carried out using a non-linear least squares estimator 7. The arameter estimates are resented in Table II. Table II Demand function: estimates Parameter Estimate Std. Error t-statistic α *** α 1 (y) α 2 (c s ) α 3 (c l ) β 0 β 1 () β 2 (y) β 3 (tem) β 4 (ner) β 5 (qual) *** *** *** *** *** *** *** *** R Source: own elaboration Parameters α 2 and α 3 show that the effectiveness of an hour of suly cut in reducing consumtion deends on the number of hours er day that the suly is interruted. When suly cuts are minor than six hours er day the reduction in consumtion er hour of cut is greater than the reduction achieved when suly interrutions are major than six hours er day. This finding allows us to suort the idea than it would be more efficient to interrut suly by means of several short eriods instead a few long ones. With shorter cuts, it would be ossible to get an aim of consumtion reduction, minimizing the total duration of interrution and users welfare losses. 7 See Greene (2002), chater 9, for details. 13

15 The (own) rice elasticity of demand of water, calculated at the arithmetic mean of rice and quantity demanded, is This result is consistent with existing studies on water demand (Arbues et al. 2003). Thus, rices are not a very effective rationing instrument in the water context. Additionally, the arameter which catures the influence of income on water demand is ositive but not statistically significant. However, as we can see high income users seem to have more ossibilities to install higher storage caacity, reducing in this way the imact of suly cuts on consumtion. So, this result shows that rationing by means of suly interrutions is a olicy which has clearly regressive effects. The remaining variables show the exected signs. Quality reductions lead to very significant reduction in consumtion. Finally, temeratures and the number of eole in household have a ositive and significant imact in water demand. Regarding welfare simulation exercise, we have based on the results of the demand estimation. Thus, we consider two alternative scenarios: the first is when there are no restrictions on the quality of water rovided (QUAL = 0), and the second when there are restrictions on quality (QUAL = 1). Additionally, we consider both short and long cuts. From the estimates, and taking the average values of the variables we calculate: a) the loss of surlus caused by the average interrution ( welfare loss associated with v ( W quality reduction ( W LQ C W ), b) the net ) and finally c) welfare loss associated with ). Table III shows the simulation results. Notice that, in order to observe the imact of several means of rationing (rices and suly interrutions), it would be necessary to comare C W with W. 14

16 Table III Welfare simulation QUAL=0 QUAL=1 c s(=223) c l(=545) c s(223) c l=(545) Q NR Q v v q W c W W LQ Source: own elaboration In the revious table and as we exected, we observe that welfare losses are higher during the eriods in which long interrutions are alied. Additionally, we observe that the main welfare losses are due to water quality reductions. Scarcity and quality roblems are shown in virtual rices values. When water quality is low, residential users would be able to ay almost twice the real rice. 6.- Conclusions Water scarcity is a cyclical roblem which water suliers have to deal with. In this research, we have focused on analyzing the efficiency and effectiveness of suly cut, one of the most usual rocedures to manage droughts. In an emirical alication, we study the effects of the interrutions imosed by the Sevillian authorities during the drought in the mid-1990s. At this resect, we have found that shorter and more frequent cuts are referred to few longer cuts. That rocedure allows getting a lanned reduction in residential water consumtion minimizing the total hours of interrution. Moreover, 15

17 we find that suly cuts are regressive in the sense that the higher the level of household income, the lower the effect of the cut. This is consistent with the fact that higher income households have easier access to water storage technologies. Additionally, regarding other rationing methods, we have obtained, as in revious studies, a low rice-elasticity value. Definitively, those are very significant findings in order to design rationing olicies in the context of water field. 16

18 References Arbués, F., M. A. García-Valiñas, and R. Martínez-Esinera (2003). Estimation of residential water demand: A state-of-the-art review. The Journal of Socio- Economics, 32, Charney, A.H. and G.C. Woodard (1984). A test of consumer demand resonse to water rices: Rely. Land Economics, 60, García-Valiñas, M.A. (2006). Analysing rationing olicies: drought and its effects on urban users welfare. Alied Economics, 38(8), Gibbons, D.C. (1986). The Economic Value of Water. Washington, D.C.: Resources for the Future. Greene, W. H. (2002). Econometric Analysis. New Jersey: Prentice Hall. INE (2006). Cifras de Población Referidas al 1/1/06. Población de Municiios or Sexo. Madrid: Instituto Nacional de Estadística. La Caixa (2000). Anuario Comercial Esañol. Barcelona: Servicio de Estudios de La Caixa. Lee, T.R., (1999). Water Management in the 21st Century: The Allocation Imerative. Cheltenham: Edward Elgar Publishing. Molina, A. (2001). El Servicio Público de Abastecimiento de Agua en Poblaciones. El Contexto Liberalizador. Valencia: Tirant lo Blanch. Moncur, J. (1987). Urban water ricing and drought management. Water Resources Research, 23, Oaluch, J. (1984). A test of consumer demand resonse to water rices: Rely. Land Economics, 60,

19 Renwick, M.E. and S.O. Archibald (1998). Demand side management olicies for residential water use: Who bears the conservation burden?. Land Economics, 74, Roibas, D., Garcia-Valiñas, M.A. and Wall, A. (2007). Measuring welfare losses from interrution and ricing as resonses to water shortages: An alication to the case of Seville. Environmental and Resource Economics, forthcoming. Tirole, J. (1988). The Theory of Industrial Organization. Cambridge: The MIT Press. Winenny, J. (1994). Managing Water as an Economic Resource. London: Routledge. Woo, C. (1994). Managing Water Suly Shortage. Interrution vs. Pricing. Journal of Public Economics, 54,