Abstract

The assessment of profit change over time and its drivers is essential to analyse firms' financial performance. This paper investigates profit change and its components for the 10 English and Welsh water and sewerage water companies over the period 1991–2008 and for three regulatory sub-periods. Profit changes and their drivers are computed following two approaches, namely: without controlling for water and sewerage quality issues, and after decomposing the output effect into high quality and low quality output effect. In both cases, profit change is decomposed into various factors such as quantity and price effect, technical change, efficiency change, resource mix, product mix, and scale effect. Profit change over the whole period was negative, with the substantial negative price effect being the main driver, which outstripped the positive quantity effect. This negative profit change was significantly marked from 2000 to 2005 while in the sub-period 1994–2000, which covers the 1994 price review, the profit change was positive. A further decomposition illustrated the significant negative impact of the input price and scale effects on profit changes. The methodology and conclusions of this paper are of great interest for both regulators and water utilities managers to improve future performance.

Introduction

Privatization and/or regulation with an appropriate form of incentive regulation are widely considered to be appropriate policy responses (Wang, 2014). Hence, in recent decades, the water industry in many countries including the UK, The Netherlands, Portugal, Chile, Colombia, and Argentina has come to be regulated with incentive regulation (Estache & Trujillo, 2003; Conti, 2006; Mugisha, 2006; Berg & Marques, 2011). Nevertheless, the reform of the water sector is not free of debate and controversy among both academics and practitioners (Marques, 2008; Berg, 2013). This is especially important in the case of the water industry in England and Wales, because criticism has been focused on the high profits earned in the industry since privatization in 1989 (Saal & Parker, 2001).

From a policy perspective to evaluate the performance change of firms, it is essential to estimate not just profit change but also its drivers (De Stefano et al., 2014). In this sense, changes in profits over time can be attributed to changes in both productivity and prices. This is the approach followed by Saal & Parker (2001) in the UK water industry framework, which estimated total factor productivity and total price performance from 1985 to 1999. However, there are other determinants that might affect profit changes over time such as technical change, efficiency change, scale effects, and resource and product mix effects (Grifell-Tatje & Lovell, 1999).

The profit decomposition developed by Grifell-Tatje & Lovell (1999) can be applied in a regulatory framework to assess the impact of price cap regulation on the financial performance of the regulated companies (De Witte & Marques, 2012). This methodology would enable both regulators and regulated companies to better identify the sources of profit variation, and aid them in evaluating both the effectiveness of a regulatory price cap scheme and the performance of the regulated companies. Also, profit decomposition enables the regulator to identify those sources of profits that can eventually be passed along to consumers in lower output prices, such as those attributable to productivity gains (Bhatt, 2014). Moreover, this methodology can also be used by the regulated companies to identify the determinants of their profit changes and improve future performance, thereby leading to future profit gains (Correira & Marques, 2008).

Since its introduction, the profit decomposition suggested by Grifell-Tatje & Lovell (1999) and subsequent developments has been applied to a wide variety of organizations and services such as railroads (Lim & Lovell, 2009), post offices (Grifell-Tatje & Lovell, 2008), banks (Sahoo & Tone, 2009; Grifell-Tatje, 2011; Juo, 2014), dairy farms (Sipiläinen et al., 2014), and telecommunication services (Banker et al., 2013). However, in the framework of the water industry, to the best of our knowledge, only De,Witte & Saal (2010) have followed Grifell-Tatje & Lovell's (1999) approach to decompose the profit change of the Dutch water sector, employing Laspeyres and Paasche indicators and Free Disposal Hull techniques to adapt Grifell-Tatje and Lovell's methodology after controlling for differences in the operating environment. Thus, profit change is decomposed into several factors such as output and input prices, productivity effect, product mix, resource mix, and scale effects.

This paper is concerned with profit change and its drivers in the water and sewerage companies (WaSCs) in England and Wales during the period 1991–20081, i.e., after the privatization process. In doing so, first, it follows Griffel-Tatje & Lovell's (1999) and De Witte & Saal's (2010) approaches to decompose profits into quantity and price effect using Bennet indicators to weigh the changes in quantities and prices. Subsequently, data envelopment analysis (DEA) techniques are applied to take into account the impact of efficiency change technical change and scale effects on profit changes. It should be noted that the Dutch water industry is regulated following a sunshine approach, while the English and Welsh water sector is operated under price cap regulation. Hence, the policy implications of De Witte & Saal's (2010) study and this paper might diverge.

English and Welsh WaSCs have carried out substantial capital investment projects to improve drinking water quality and environmental standards2 (Portela et al., 2011). Hence, a second assessment of the profit change introducing quality issues has been performed. As these substantial drinking water quality and sewerage treatment improvements have been in response to increasingly stringent environmental regulation, including European Union directives (Maziotis et al., 2009), it is assumed that quality improvements are exogenously determined (Saal & Parker, 2000). The assessment of the profit change controlling for quality allows consideration of whether the inclusion of quality affects the drivers of the profit change. Finally, taking into account that the period evaluated involves three regulatory sub-periods (1994–2000; 2000–2005; 2005–2008), our study also provides some insights into the relation between profit changes and the regulatory cycle.

This paper contributes to the current strand of literature in the field of water companies' performance measurement by comparing the drivers of profit change with and without controlling for water and sewerage quality issues. In the UK context, this paper is pioneering in decomposing profit change into several factors such as price effect, efficiency change, technical change, resource mix, product mix, and scale effect. Moreover, the paper uses more recent data than previous studies in this area.

From a policy perspective, the results and conclusions of this study are of great interest for both regulators and utility managers. In countries or regions where price cap schemes are employed in conjunction with a system of yardstick competitions, information about the profit change of water companies would be very useful in assisting price reviews. Moreover, the decomposition of profit change into several drivers allows water utilities' managers to identify the main factors on which they should act to increase the profits of the company.

This paper unfolds as follows. Section 2 discusses the concept of distance functions. It includes an analysis of the decomposition of profits into its components and the DEA technique employed in order to estimate the components of the profit decomposition without and with adjustments for quality. Section 3 presents the data that are used in our study. Section 4 discusses the empirical results, while the last section concludes.

Methodology

2.1. Profit decomposition without controlling for quality

The production technology at each period t is defined as the set that includes all feasible output–inputs correspondences. The inputs are represented by a positive input quantity vector where N denotes the total number of inputs that a company uses in order to produce a vector of non-negative outputs, where M denotes the total number of outputs. Let us assume that we have a positive vector of input prices and a positive vector of output prices . The production technology for period t is then represented as follows: 
formula
(1)
Let also the input set, , represent the set of all input vectors that can produce a given output vector at period t as follows: 
formula
(2)
Shephard (1970) introduced the input distance function to provide a functional representation of production technology. The input distance function is defined as a minimal proportional reduction of the input vector given an output vector at each period t and is given by the following: 
formula
(3)
for and for .
Let us also define the output set, , which represents the set of all output vectors, , that can be produced using the input vector, , in period : 
formula
(4)

represents the efficient input boundary that is the locus of minimum input levels needed to produce a given level of output in period t.

Shephard's (1970) output distance function provides another functional representation of production technology. The output distance function is defined as a maximal proportional expansion of the output vector given an input vector at each period t and is given by the following: 
formula
(5)
for and for .

The distance functions, being radial distance measures, provide the tools with which we will recover the unobserved quantity vectors that we need for profit decomposition.

To estimate the contributions of quantity and price effect to profit change, De Witte & Saal's (2010) approach is followed. It provides an input-oriented profit decomposition between two time periods t and using Bennet indicators, average prices and quantities as weights. Let a company's profit in period t, , be defined as a difference between its total revenues and total costs, . Using Bennet indicators, , , profit change between periods t and , is decomposed into two factors, namely quantity effect (QE) and price effect, as follows: 
formula
(6)

The QE captures the contribution to profit change from a change in output production and input usage, while the price effect shows the contribution to profit change from a change in output and input prices. The QE shows that profits may increase due to a rise in output production in excess of the corresponding input rise, while the price effect shows that profits may also rise due to an increase in output prices in excess of the rise in input prices. The decomposition of profits into a quantity and price effect involves only observed quantity and price data.

As shown in Figure 1, the QE can in turn be decomposed into other effects such as the productivity effect (PROE) and an activity effect (AE) as follows: 
formula
(7)
Fig. 1.

Decomposition of the profit change into its components. Source: Own elaboration based on Grifell-Tatje & Lovell (1999).

Fig. 1.

Decomposition of the profit change into its components. Source: Own elaboration based on Grifell-Tatje & Lovell (1999).

This decomposition is depicted in Figure 2. The QE as decomposed in Equation (7) makes use of the observed quantities to and of the unobserved quantities . According to Figure 2, and denote the efficient input level that the unit could have used in periods t and , respectively, to secure out keeping to the input mix of , while represents the efficient input level that the unit could have used in period to secure out keeping to the input mix of .

Fig. 2.

Productivity effect. Source: Own elaboration based on Grifell-Tatje & Lovell (1999).

Fig. 2.

Productivity effect. Source: Own elaboration based on Grifell-Tatje & Lovell (1999).

The productivity effect in Equation (7) compares the distance from to in period t with the distance from to in period . The difference in these two distances reflects the productivity change of the unit, as it captures how much closer or further from the ‘fixed’ efficient boundary of period the unit has moved over time. If , the contribution to profit change is positive, whereas if , the contribution to profit change is negative.

The AE in Equation (7) measures the changes in the scale and scope of the activities of a company. When is positive, it reflects a rise in output over time, while a negative reflects a fall in the efficient level of input needed to secure the output. Thus, both the output and the input differences in this case lead, respectively, to positive contributions to profit change between periods t and .

Finally, in a third stage decomposition, the productivity effect in Equation (7) can be further decomposed into an efficiency change effect and technical change effect. Moreover, the AE can be further decomposed into a resource mix, output mix, and scale effect. Figures 2 and 3 depict the decomposition of the productivity effect and AE, which we now elaborate upon. 
formula
(8)

The technical change effect is measured by the distance from to . As can be seen in Figure 2, this difference reflects the distance between the efficient boundaries in periods t and , controlling for output level. Technical improvement occurs when . Such an improvement in the efficient boundary from t to has a positive effect on profit change from t to , whereas with technical regress, there will be a negative impact on profit change.

Moving to the efficiency change effect, the distance from to reflects the inefficiency of the firm in period t, and similarly, the distance from to reflects the inefficiency of the firm in period . Thus, as illustrated in Equation (8), a decline in the input price weighted cost of inefficiency in period , relative to the equivalent cost in period t, has a positive impact on profit change. In contrast, a rise in the input price weighted cost of inefficiency in period relative to that in period t would have a negative impact on profit change.

The AE in Equation (7) can be further decomposed as follows: 
formula
(9)
The resource mix effect captures the impact on profits due to the change in the mix of inputs between periods t and while keeping the output at the period level and also retaining efficiency in production (see Figure 2). When is positive, the change in resource mix reflects a movement of input usage to one which reduces costs, thereby improving allocative efficiency. Similarly, we can infer from Figure 3 the product mix effect as the change in output mix from to . Note that reflects the product mix of period t, but its level is that resulting from using the efficient input level in period to secure the product mix of period t.
Fig. 3.

Activity effect. Source: Own elaboration based on Grifell-Tatje & Lovell (1999).

Fig. 3.

Activity effect. Source: Own elaboration based on Grifell-Tatje & Lovell (1999).

Finally, the scale effect consists of two components, the input scale effect and the output scale effect, thereby capturing the impact of scale change on the firm's profitability. From Figure 2, it is noted that to produce efficiently the output of period t, , using the best practice technology available in period , the input level needed is . In contrast, when outputs change from to , while keeping the input mix and the technology constant to that of period , the input required is . The difference between and when positive means that the efficient input level needed in constant technology has dropped as output changed from periods t to and this has a positive impact on profit. As and have the same mix, their difference simply reflects the difference in their scale size. In a similar manner, and have the same mix, as can be seen in Figure 3, and their difference reflects the difference in their scale size. and are efficient output levels on technology using, respectively, input levels and already defined in Figure 2.

The second and third stage of the above profit decomposition requires the computation of the unobserved quantities . These unobserved input and output quantities can be estimated using DEA by means of the input and output distance functions (Equations (3) and (5)) as follows: 
formula
(10)
Let J, N, M, and T denote, respectively, the total number of firms, inputs, outputs, and time periods in the sample. Let denote a scalar which represents the proportional contraction of the input vector, given the output vector, and denote a scalar which represents the proportional expansion of the output vector, given the input vector. Let and denote the output vector and the input vector, respectively, for the jth firm in the tth period . Let and denote, respectively, the output matrix and the input matrix in period t, containing the data for all the firms in the t-th period. The notation for period is defined similarly. We use the additional constraint to allow for variable returns to scale technology. The reference technology for our DEA models is the sequential DEA technology which is defined in Section 3 (Sample and data description). Sequential technology assumes that in any period t the technology of the previous periods remains feasible. By definition, this technology does not allow for regress. Thus, in period t, the unobserved quantity can be computed by the following linear programming problem: 
formula
(11)

The variables , whose optimal values are to be determined by the above model, lead to the estimate of the proportional reduction in that would locate on the efficient frontier within the sequential technology to period t. The unobserved quantity for a firm having input output set is thus , and it is computed for each firm in the sample in period t.

In an analogous manner, is computed using data for all the firms in the period with technology of period, whereas is computed using data for all the firms in the t period with technology of period.

The unobserved quantity can be computed by the following linear programming problem: 
formula
(12)
The unobserved quantity is computed as for each firm in the sample in period t. The unobserved quantity can be computed using Equation (12) and the following linear programming problem: 
formula
(13)

The unobserved quantity is computed as for each firm in the sample in period t.

2.2. Profit decomposition controlling for quality

As earlier (Section 2.1), the inputs are represented by a positive input quantity vector where N denotes the total number of resources, and the positive vector of input prices can be defined as However, the positive vector of output quantities where M denotes the total number of outputs is now separated into a non-negative vector of output for high quality and a non-negative vector of output for low quality where H and L denote the total number of outputs for high and low quality, respectively, and we assume that and that more inputs are required to produce a given amount of high quality output than to produce the same amount of low quality output. However, the above modification in the quality of output cannot be applied in the case of outputs prices. In the UK water industry, all customers of a given water firm effectively pay the same price for water services regardless of output quality, as regulated water prices do not differentiate between quality of output. Moreover, given this regulatory practice, it is unsurprising that, while total turnover data are available separately for water and sewerage services, it is not disaggregated by quality of service. As a result, we do not in practice have different prices for high and low quality water and sewerage output types, even though we can observe quantity data reflecting differences in output quality.

It should be noted that additional input types may be needed for producing higher quality output that are not necessary for producing output of lower quality. For example, different facilities and chemicals are needed at different stages of sewerage treatment. Prices for the different types of resources used for outputs of different quality may also differ. However, according to previous studies in the framework of the UK water industry (Saal & Parker, 2000, 2001, 2006; Saal et al., 2007), it is assumed that no additional input types are needed for producing higher quality output and that prices of inputs are independent of the mix of output quality. Nevertheless, in our empirical application in the linear programming models, we imposed the weight restriction that the production of high quality output is at least as resource-intensive as the same quantity of output of low quality.

Given the single output price and , profit change decomposes into QE and price effect as follows: 
formula
(6′)

Since and the decomposition of profits into quantity and price effect is calculated using observed quantities and prices, the results from the first stage of the profit decomposition in Equation (6′) will in aggregate be exactly the same as in Equation (6). The one distinction is that the QE now captures the contribution to profit changes from a change in output production of high and low quality and input usage, using the observed output prices as weights to evaluate the changes in the high and low quality output effect.

The decomposition of the QE into a productivity effect and AE in Equation (7) becomes: 
formula
(7′)
The results for the productivity effect and AE in Equation (7′) will differ from those calculated in Equation (7), since it uses the unobserved input quantities , where estimates with DEA will need to include two output vectors, and , instead of the aggregate vector Y, in the linear programming models. In an analogous manner, the results from the further decomposition of the productivity effect and AE into their components will also differ from those obtained in Equations (8) and (9), since the recovery of the three unobserved input and output quantities needs to include two output vectors, and , instead of the aggregate vector Y. For instance, the decomposition of the AE into resource mix effect, product mix effect, and scale effect will become: 
formula
(9′)
The scale effect now captures the change in the efficient output levels for high and low output quality given efficient input usage, whereas the resource mix effect now reflects changes in the efficient mix of input usage to secure output of high and low quality. It is worth noting that the product mix effect does not reflect changes in the product mix for high and low quality, but only changes in the aggregate non-quality differentiated mix of outputs due to the assumption of quality undifferentiated output prices.

Sample and data description

The empirical application is focused on the assessment of profit change during the period 1991–2008 for the 10 English and Welsh WaSCs. The source of the data essentially comes from ‘June Returns for the Water and Sewerage Industries in England and Wales' published by Ofwat3 each year on its webpage. Water connected properties and sewerage connected properties are the outputs.

Further, three inputs, namely capital stock, full-time equivalent employees and other costs in constant pounds, are used in the analysis. Following Saal & Parker (2001) and Maziotis et al. (2009, 2012), the capital stock measure was based on the inflation adjusted modern equivalent asset (MEA) estimates of the replacement cost of physical assets in the companies' regulatory accounts. However, due to periodic revaluations of these replacement cost values, real net investment was used. Moreover, the resulting year-ending and year-beginning estimates were averaged to provide a more accurate estimation of the average physical capital stock available to the companies in a given year. Subsequently, a user-cost of capital approach is employed to calculate total capital costs as the sum of the opportunity cost of invested capital and capital depreciation relative to the MEA asset values. Then, the price of physical capital is constructed as the user-cost of capital divided by the above MEA-based measure of physical capital stocks (Maziotis et al., 2009, 2012). Regarding the second input, the cost per unit of labour was calculated as the ratio of total labour costs to the average number of full-time equivalent employees. Moreover, other costs input (in nominal terms) was defined as the difference between operating costs and total labour costs. Given the absence of data allowing a more refined breakdown of other costs, we have employed the UK price index for materials and fuel purchased in purification and distribution of water as the price index for other costs, and simply deflated nominal other costs by this measure to obtain a proxy for real usage of other inputs.

Water and sewerage output prices were calculated as the ratio of the appropriate turnover in nominal terms, as available in Ofwat's regulatory returns. Economic profits were calculated as the difference between turnover and calculated economic costs.

Regarding the quality-adjusted analysis, data from the DWI's annual reports for the calendar years ending 1991–20074 were used. Following Saal & Parker (2001), high drinking water quality was defined as the average percentage of each WaSC's water supply zones that were compliant with six key water quality parameters. On the other hand, low drinking water quality was defined as the average percentage of each WaSC's water supply zones that were not compliant with the same six key water quality parameters. These quality parameters reflect how well treatment works and distribution systems are operated and maintained (Ofwat, 2006). The water output for high quality was calculated as the product of water connected properties and high drinking water quality . Analogously, the water output for low quality was defined as the product of water connected properties and low drinking water quality . The water output price was the same for high and low quality, and was defined as the ratio of water total turnover in nominal terms to the sum of water output for high and low quality.

Sewerage quality was defined based on data regarding the percentage of the connected population for which wastewater receives various types of treatment (non-treatment, primary, secondary, or advanced). Data were taken from Waterfacts for the period 1990–1996 and the companies' regulatory returns for 1997–2008. High sewerage treatment quality was defined as the percentage of the connected population receiving at least secondary or higher wastewater treatment, while low sewerage treatment quality was defined as the percentage of the connected population receiving non- or primary wastewater treatment. The sewerage output for high quality was calculated as the product of sewerage connected properties and the percentage of the connected population receiving at least secondary or higher wastewater treatment . The sewerage output for low quality was calculated as the product of sewerage connected properties and the percentage of the connected population receiving non- or primary wastewater treatment . The sewerage output price was the same for high and low quality, and was defined as the ratio of sewerage total turnover in nominal terms to the sum of sewerage output for high and low quality.

Table 1 shows the aggregate data for the sample, with all the data expressed in real 2008 prices.

Table 1.

Aggregate profits, revenues, costs, outputs, and inputs (expressed in real 2008 prices).

Variables Units 1994 2000 2001 2005 2006 2008 
Economic profit 106£ 859.1 1,299.7 −186.0 −227.1 386.4 −253.5 
Revenues 106£ 7,126.4 7,908.2 7,162.9 7,491.9 8,198.2 8,494.6 
Total economic costs 106£ 6,267.3 6,608.4 7,349.0 7,718.9 7,811.8 8,748.0 
Water connected properties 10316,665 18,304 19,302 19,821 19,972 20,061 
High quality-adjusted water connected properties 10315,101 17,237 18,412 19,083 19,297 19,442 
Low quality-adjusted water connected properties 1031,564 1,067 890 738 676 619 
Sewerage connected properties 10321,298 22,123 22,274 23,017 23,456 23,795 
High quality-adjusted sewerage connected properties 10316,963 19,239 20,939 22,647 23,186 23,072 
Low quality-adjusted sewerage connected properties 1034,335 2,884 1,335 370 270 723 
Capital 106£ 192,295 206,597 208,168 213,253 214,362 216,918 
Number of employees Full-time equivalent 38,125 29,685 27,854 27,197 27,554 29,524 
Other inputs 106£ 999.5 970.1 958.6 824.6 819.3 781.4 
Average price for a quality-adjusted water connected property £ 219.01 198.75 176.9 178.06 193.87 204.35 
Average price for a quality-adjusted sewerage connected property £ 229.87 229.5 175.07 168.01 178.43 185.7 
Price for capital £ 0.017 0.019 0.023 0.024 0.024 0.028 
Price for labour 103£ 32.17 33.78 33.46 37.63 37.92 36.85 
Price of other inputs Price index 0.740 0.767 0.762 0.889 0.957 1.000 
Variables Units 1994 2000 2001 2005 2006 2008 
Economic profit 106£ 859.1 1,299.7 −186.0 −227.1 386.4 −253.5 
Revenues 106£ 7,126.4 7,908.2 7,162.9 7,491.9 8,198.2 8,494.6 
Total economic costs 106£ 6,267.3 6,608.4 7,349.0 7,718.9 7,811.8 8,748.0 
Water connected properties 10316,665 18,304 19,302 19,821 19,972 20,061 
High quality-adjusted water connected properties 10315,101 17,237 18,412 19,083 19,297 19,442 
Low quality-adjusted water connected properties 1031,564 1,067 890 738 676 619 
Sewerage connected properties 10321,298 22,123 22,274 23,017 23,456 23,795 
High quality-adjusted sewerage connected properties 10316,963 19,239 20,939 22,647 23,186 23,072 
Low quality-adjusted sewerage connected properties 1034,335 2,884 1,335 370 270 723 
Capital 106£ 192,295 206,597 208,168 213,253 214,362 216,918 
Number of employees Full-time equivalent 38,125 29,685 27,854 27,197 27,554 29,524 
Other inputs 106£ 999.5 970.1 958.6 824.6 819.3 781.4 
Average price for a quality-adjusted water connected property £ 219.01 198.75 176.9 178.06 193.87 204.35 
Average price for a quality-adjusted sewerage connected property £ 229.87 229.5 175.07 168.01 178.43 185.7 
Price for capital £ 0.017 0.019 0.023 0.024 0.024 0.028 
Price for labour 103£ 32.17 33.78 33.46 37.63 37.92 36.85 
Price of other inputs Price index 0.740 0.767 0.762 0.889 0.957 1.000 

Since the sample data includes 10 WaSCs, the estimation with DEA was modified in order to deal with the small number of observations each year. Drawing on a foregoing and a sequential technology (Tulkens & Vanden Eeckant, 1995), the reference technology for the DEA models was as follows. Data from 1991 to 1994 were pooled in order to increase the number of observations from 10 to 40. The observations from these years were used as a cross section to construct the reference technology, which was referred to as the t = 1994 frontier. The second sub-panel contains the period 1991–1995, and the frontier constructed using these data was considered as the t + 1 = 1995 frontier. The same procedure was repeated until the last sub-panel, which includes the period 1991–2008.

Results and discussion

4.1. Quality-unadjusted profit change decomposition

Table 2 displays cumulative profit change and its drivers for the entire 1994–2008 period and the regulatory sub-periods 1994–2000, 2000–2005 and 2005–2008, without making any allowances for any differences in the quality of outputs. Over the entire 1994–2008 period, the QE, efficiency change, resource mix, and scale effect contributed positively to profit changes, while the price effect, technical change, and product mix effect contributed negatively to profit changes. Focusing on aggregate profit change, profits reduced by £1,112.6 million over the period 1994–2008, which was the result of significant aggregate profit decrease during the years 2000–2008 and significant aggregate profit increase during the years 1994–2000.

Table 2.

Cumulative profit change and its decomposition (2008 pounds, millions).

 1994–2008 1994–2000 2000–2005 2005–2008 
Profit change −1,112.6 440.6 −1,526.8 −26.4 
Quantity effect 1,335.7 538.8 676.4 120.5 
 Output effect 1,080.4 482.6 413.7 184.1 
 Input effect 255.3 56.2 262.6 −63.6 
  Productivity effect 1,155.9 589.7 506.0 60.2 
   Technical change 1,041.5 609.3 348.6 83.6 
   Efficiency change 114.4 −19.6 157.4 −23.4 
  Activity effect 179.8 −50.9 170.4 60.3 
   Resource mix 939.2 147.5 355.7 436.0 
   Product mix −2.1 47.1 −90.5 41.3 
   Scale effect −757.2 −245.5 −94.8 −417.0 
Price effect −2,448.3 −98.1 −2,203.2 −146.9 
 Output price effect 287.7 299.2 −830.0 818.6 
 Input price effect −2,736.00 −397.33 −1,373.14 −965.53 
 1994–2008 1994–2000 2000–2005 2005–2008 
Profit change −1,112.6 440.6 −1,526.8 −26.4 
Quantity effect 1,335.7 538.8 676.4 120.5 
 Output effect 1,080.4 482.6 413.7 184.1 
 Input effect 255.3 56.2 262.6 −63.6 
  Productivity effect 1,155.9 589.7 506.0 60.2 
   Technical change 1,041.5 609.3 348.6 83.6 
   Efficiency change 114.4 −19.6 157.4 −23.4 
  Activity effect 179.8 −50.9 170.4 60.3 
   Resource mix 939.2 147.5 355.7 436.0 
   Product mix −2.1 47.1 −90.5 41.3 
   Scale effect −757.2 −245.5 −94.8 −417.0 
Price effect −2,448.3 −98.1 −2,203.2 −146.9 
 Output price effect 287.7 299.2 −830.0 818.6 
 Input price effect −2,736.00 −397.33 −1,373.14 −965.53 

Looking at the first stage of profit decomposition, we conclude that over the entire period, the negative effect on cumulative profit change was attributed to a significant negative price effect, which outstripped the positive QE. Within this first stage of profit decomposition, it is illustrated that during the years 1994–2008, output prices increased profits; however, greater increases in input prices contributed negatively to profit changes, resulting in the overall negative price effect. Focusing on the sub-periods of our sample, we conclude that during the years 1994–2000, covering the end of the first price review after privatization and the entire 1995–2000 period covered by the 1994 price review, there was a small increase in output prices contributing positively to profit changes. However, substantial increases in input prices counteracted this. Furthermore, the dramatically tightened 1999 price review obliged companies to reduce their output prices, and continuing increases in input prices resulted in a negative overall price effect. During the years 2005–2008, output prices increased significantly, providing evidence that the 2004 price review was relatively loose and thereby contributed positively to profit changes.

Looking at the second stage decomposition, Table 2 shows that the positive QE over the entire period was attributed to a significant positive productivity effect and a small but positive AE. Almost the entire productivity effect can be explained by technical change, while the contribution of increased efficiency was much more modest. Focusing on the decomposition of the AE, it is concluded that in aggregate, the positive AE was exclusively explained by a substantial positive resource mix effect, which was unfortunately offset by a very substantial negative scale effect as well as a quite small negative product mix effect. The resource mix effect increased significantly over the entire period and especially during the years 2000–2005 and 2005–2008, suggesting movement to a more cost efficient allocation of resources more in line with relative factor prices. Moreover, the scale effect was negative for all sub-periods evaluated. This finding suggests that mergers were not profitable for WaSCs, which is in contrast to the Cave Review (2009) recommendations, which suggested further mergers in the UK water and sewerage industry. Changes in the mix of outputs, specifically the production of more output for water services than sewerage services, increased profits during the years 1994–2000 and 2005–2008 but decreased profits significantly during the years 2000–2005.

From a policy perspective and relating the results from the decomposition of profits into several factors with the regulatory cycle, we conclude that during the years 1994–2000, when price caps were tightened after the 1994 price review, profits increased. This increase in aggregate profitability was attributed to the positive cumulative QE and still increasing output prices, which largely offset substantial increases in input prices. There were also significant improvements in productivity mainly attributable to technical change, indicating that the most productive companies significantly improved their performance. During the years 2000–2005, when profits substantially decreased, the cumulative impact of price effect is captured by a significant reduction in output prices due to the tightened 1999 price review and a high increase in input prices that both offset the positive QE. However, there were still substantial productivity improvements attributed to both technical change and increased efficiency, indicating that both the most productive and less productive firms had strong incentives to improve their productivity in order to regain economic profitability. Finally, during the years 2005–2008, when profits reduced very slightly, this was explained by a positive cumulative impact of the QE, and substantial gains in output prices, the combined effect of which was nonetheless almost completely offset by large increases in input prices.

4.2. Quality-adjusted profit change decomposition

This section presents the results from cumulative profit change and its components when differences in the quality of output are considered. As discussed in Section 2.2, given the lack of distinct price data for low and high quality outputs in the first stage of profit decomposition (quantity and price effect), the quality-adjusted and quality-unadjusted results are the same, except for the detail provided with respect to changes of quality on the aggregate output effect. Thus, the results indicate that over the whole period, WaSCs moved to the production of more high quality output while reducing the production of low quality output. Nevertheless, as our following discussion illustrates, accounting for differences in output quality, and the resulting impact on efficient input quantities, does in fact alter the decomposition of the QE, thereby improving our understanding of the sources of profit change.

Thus, in the quality-adjusted estimation provided in Table 3, it remains the case that the positive QE over the entire 1994–2008 period can be predominantly attributed to the significant positive productivity effect. However, the quality-adjusted results suggest a relatively greater role for the AE, which increases from £179.8 to £246.2 million relative to the quality-unadjusted results. This result suggests a greater weight placed on the reorganization of activities as opposed to productivity effects. This change in results is consistent with allowing for changes in output quality in the model over a period when substantial quality changes occurred.

Table 3.

Cumulative high and low quality-adjusted profit change and its decomposition (2008 pounds, millions).

 1994–2008 1994–2000 2000–2005 2005–2008 
Profit change −1,112.6 440.6 −1,526.8 −26.4 
Quantity effect 1,335.7 538.8 676.4 120.5 
 Output effect 1,080.4 482.6 413.7 184.1 
  High quality output effect 2,067.1 902.3 1,015.5 149.3 
  Low quality output effect −986.6 −419.8 −601.7 34.9 
 Input effect 255.3 56.2 262.6 −63.6 
  Productivity effect 1,089.5 563.5 457.4 68.6 
   Technical change 989.4 556.1 321.6 111.8 
   Efficiency change 100.0 7.4 135.8 −43.2 
  Activity effect 246.2 −24.7 219.0 52.0 
   Resource mix 1,176.1 275.6 520.8 379.7 
   Product mix 30.4 −60.9 81.8 9.5 
   Scale effect −960.3 −239.5 −383.5 −337.3 
Price effect −2,448.3 −98.1 −2,203.2 −146.9 
 Output price effect 287.7 299.2 −830.0 818.6 
 Input price effect −2,736.00 −397.33 −1,373.14 −965.53 
 1994–2008 1994–2000 2000–2005 2005–2008 
Profit change −1,112.6 440.6 −1,526.8 −26.4 
Quantity effect 1,335.7 538.8 676.4 120.5 
 Output effect 1,080.4 482.6 413.7 184.1 
  High quality output effect 2,067.1 902.3 1,015.5 149.3 
  Low quality output effect −986.6 −419.8 −601.7 34.9 
 Input effect 255.3 56.2 262.6 −63.6 
  Productivity effect 1,089.5 563.5 457.4 68.6 
   Technical change 989.4 556.1 321.6 111.8 
   Efficiency change 100.0 7.4 135.8 −43.2 
  Activity effect 246.2 −24.7 219.0 52.0 
   Resource mix 1,176.1 275.6 520.8 379.7 
   Product mix 30.4 −60.9 81.8 9.5 
   Scale effect −960.3 −239.5 −383.5 −337.3 
Price effect −2,448.3 −98.1 −2,203.2 −146.9 
 Output price effect 287.7 299.2 −830.0 818.6 
 Input price effect −2,736.00 −397.33 −1,373.14 −965.53 

Delving into the components of the productivity effect further reveals that while the magnitude of technical change's contribution declines in the quality-adjusted models, it still remained the major determinant of the productivity effect over 1994–2008, with efficiency change still contributing very little to profit change. Focusing on the regulatory sub-periods, technical change was large and positive during the years 1994–2000 and 2000–2005, whereas for the years 2005–2008, although it was positive, it was much more moderated. In contrast, efficiency change was positive during the years 1994–2000 and 2000–2005, while it became negative during the years 2005–2008. Thus, the impression that efficiency change was relatively unimportant in improving firm performance, as well as a dampening of efficiency incentives after the 2004 price review, is both reinforced in the quality-adjusted results.

The relatively small increase in the overall AE after accounting for output quality actually disguises much more substantial changes in its components. Thus, for the 1994–2008 period, the resource mix and product mix effects, respectively, increase from £939.2 and −£2.1 to £1,176.1 and £30.4 million, while in contrast, the scale effect becomes even more negative as it declines from −£757.2 to −£960.3 million. Given the lack of differentiated quality output prices, these results therefore suggest that with controls for quality, the estimated benefits of changing to a more capital intensive input mix are more substantial. In contrast, the negative effects on input usage and hence costs attributable to increases in scale are estimated to be even more substantial with controls for quality.

Focusing on the direct interpretation of the quality-adjusted AE and its decomposition reveals that the aggregate positive AE is mainly explained by high positive resource mix and small product mix effects, which outstripped a very large and substantial negative scale effect. The resource mix effect contributed significantly to profit change over the entire period and especially after 2000, indicating that there was a steady shift to a more capital intensive resource allocation that was more cost-effective given observed input prices. However, the scale effect did not lower costs and reduced profits significantly during each of the sub-periods detailed in Table 3. Thus, our results suggest that the substantial savings attributed to a more cost-efficient allocation of resources, as measured by the resource mix effect, were almost entirely lost because of the excessive scale of the WaSCs and several mergers during the period when the WaSCs absorbed water-only companies. This result is consistent with previous evidence with regard to the presence of diseconomies of scale for the WaSCs (Saal & Parker, 2000; Stone & Webster Consultants, 2004; Saal et al., 2007; Bottaso & Conti, 2009).

Conclusions

The assessment of profit change over time and its drivers is essential to analyse a firms financial performance. In a regulatory framework such as the English and Welsh water sector, the evaluation of profit change also provides essential information about the impact of price cap regulation in the performance of water companies.

This paper estimates profit change and its components for the 10 English and Welsh WaSCs over the period 1991–2008 following two approaches: (i) without controlling water and sewerage quality issues and (ii) introducing quality variables by decomposing the output effect into high and low quality output effects. In both cases, profit change is decomposed into various factors such as quantity and price effect, technical change, efficiency change, resource mix, product mix, and scale effects.

The main findings of this study are as follows. First, both quality-unadjusted and quality-adjusted results evidence that over the whole period the main source of negative profit change was driven by the substantial negative price effect, which outstripped the positive QE. Second, further decomposition of profit change suggests that the WaSCs won substantial savings through labour substitution and technical change, but most of these gains were lost due to the negative effects associated with increasing scale. Third, quality-based decomposition of the output vector reveals superior unbiased estimates of the underlying technology and costs of a firm even with the assumption of undifferentiated output prices.

From a policy perspective, this study has significant implications for the regulated UK water and sewerage industry, which are summarized as follows. First, the substantial capital investment programmes carried output by the WaSCs since privatization appear to have led to the production of higher quality outputs. Second, significant productivity improvements, which contributed positively to profit changes, were mainly attributed to technical change, whereas gains in efficiency were small. Third, a substantial resource mix effect implies improvements in the cost-efficient allocation of resources by substituting labour with capital. Finally, the scale effect was negative over the whole period, which suggests that mergers were not profitable for WaSCs.

We strongly believe that the above findings are of great importance for researchers and policy makers, as they will allow further analysis exploring scale and scope economies in the water industry, determination of its most economically efficient structure, and the influence of such changes on the profitability and performance of the industry.

Acknowledgements

The authors would like to express their gratitude for the support of the Economic and Social Science Research Council as well as the Office of Water Services (Ofwat). María Molinos-Senante would like to thank Generalitat Valenciana (APOSTD/2013/110) for financial support.

1

In August 2009, as part of the wider process of further developing competition in the water industry, Ofwat published new accounting requirements for 2009–2010, which could potentially lead to formal accounting separation of the water and sewerage supply chain. Therefore, this did not allow us to collect sufficient and consistent data to accurately assess the performance of WaSCs in subsequent years.

2

Quality is policed by the Drinking Water Inspectorate (DWI) and the Environment Agency.

3

Ofwat is the Water Services Authority in England and Wales. More information can be found at http://www.ofwat.gov.uk/.

4

The DWI provides quality data based on calendar years, while all other information employed in this paper is based on fiscal years ending 31st March. We note this inconsistency in the data, but emphasize that the reported years overlap each other for 9 months. Thus, the year-end to year-end estimates of quality change obtained from the DWI data provide consistent estimates of quality change by the water companies, at a fixed point 9 months into each fiscal year.

References

References
Banker
R.
,
Ca
Z.
,
Menon
N. M.
&
Mudambi
R.
,
2013
.
The red queen in action: the longitudinal effects of capital investments in the mobile telecommunications sector
.
Industrial and Corporate Change
22
(
5
),
dts036
,
1195
1228
.
Bottasso
A.
&
Conti
M.
,
2009
.
Price cap regulation and the ratchet effect: a generalised index approach
.
Journal of Productivity Analysis
32
,
191
201
.
Cave
M.
,
2009
.
Independent Review of Competition and Innovation in Water Markets: Final Report
.
Report Prepared for and Published by Ofwat
.
Conti
M.
,
2006
.
Reliability of point efficiency estimates in regulatory applications. Evidence from the English and Welsh water industry
.
International Journal of Regulation and Governance
6
(
1
),
21
41
.
De Stefano
L.
,
Svendsen
M.
,
Giordano
M.
,
Steel
B. S.
,
Brown
B.
&
Wolf
A. T.
,
2014
.
Water governance benchmarking: concepts and approach framework as applied to Middle East and North Africa countries
.
Water Policy
16
(
6
),
1121
1139
.
Estache
A.
&
Trujillo
L.
,
2003
.
Efficiency effects of ‘privatization’ in Argentina's water and sanitation services
.
Water Policy
5
(
4
),
369
380
.
Grifell-Tatje
E.
&
Lovell
C. A. K.
,
1999
.
Profits and productivity
.
Management Science
45
(
9
),
1177
1193
.
Grifell-Tatjé
E.
&
Lovell
C. A. K.
,
2008
.
Productivity at the post: its drivers and its distributions
.
Journal of Regulatory Economics
33
,
133
158
.
Juo
J.-C.
,
2014
.
Decomposing the change in profit of Taiwanese banks: incorporating risk
.
Journal of Productivity Analysis
41
(
2
),
247
262
.
Lim
S. H.
&
Lovell
C. A. K.
,
2009
.
Profits and productivity of U.S. class I railroads
.
Managerial and Decision Economics
30
(
7
),
423
442
.
Maziotis
A.
,
Saal
D. S.
&
Thanassoulis
E.
,
2009
.
Regulatory Price Performance, Excess Cost Indexes and Profitability: How Effective is Price Cap Regulation in the Water Industry?
Aston Business School Working Papers, RP 0920
.
Maziotis
A.
,
Saal
D. S.
&
Thanassoulis
E.
,
2012
.
Profit, Productivity and Price Changes in the English and Welsh Water and Sewerage Companies
.
FEEM Working Paper, Nota di Lavoro No. 84
.
Ofwat
,
,
2006
.
Report on the Levels of Service for the Water Industry in England and Wales
.
Office of Water Services
,
Birmingham
.
Portela
M. C. A. S.
,
Thanassoulis
E.
,
Horncastle
A.
&
Maugg
T.
,
2011
.
Productivity change in the water industry in England and Wales: application of the meta-Malmquist index
.
Journal of the Operational Research Society
62
(
12
),
2173
2188
.
Saal
D.
,
Parker
D.
,
2006
.
Assessing the performance of water operations in the English and Welsh water industry: a lesson in the implications of inappropriately assuming a common frontier
. In:
Performance Measurement and Regulation of Network Utilities
.
Ceolli
T.
&
Lawrence
D.
(eds).
Edward Elgar
,
Cheltenham
.
Sahoo
B. K.
&
Tone
K.
,
2009
.
Radial and non-radial decompositions of profit change: with an application to Indian banking
.
European Journal of Operational Research
196
,
1130
1146
.
Shephard
R. W.
,
1970
.
Theory of Cost and Production Functions.
Princeton University Press
,
Princeton
.
Sipiläinen
T.
,
Kumbhakar
S. C.
&
Lien
G.
,
2014
.
Performance of dairy farms in Finland and Norway from 1991 to 2008
.
European Review of Agricultural Economics
41
(
1
),
63
86
.
Stone and Webster Consultants
2004
.
Investigation into Evidence for Economies of Scale in the Water and Sewerage Industry in England and Wales: Final Report
.
Report Prepared for and Published by
Ofwat
.
Tulkens
H.
&
Vanden Eeckaut
P.
,
1995
.
Non-parametric efficiency, progress and regress measures for panel data: methodological aspects
.
European Journal of Operational Research
80
,
474
499
.
Wang
J.
,
2014
.
Analysis on economic values and economic significance under macroeconomic regulation and control
.
WIT Transactions on Information and Communication Technologies
49
,
1375
1379
.