Assessing the Size and Scope of Labor Restructuring

BENCHMARKING

Benchmarking is an important mechanism for identifying the potential for labor productivity improvements (box 3.6). Making good comparisons can be difficult, but there are several sources of information. In addition the process of benchmarking will help identify problem areas in terms of overstaffing and opportunities for improving labor productivity.

What Are Benchmarks?

Benchmarks are fixed pieces of information that can be used to make comparisons with other similar fixed pieces of information. Labor benchmarks are not only used as a one-off activity for work force restructuring but also as a tool for continuously monitoring and improving performance and competitiveness. In practice it is the process of undertaking benchmarking that generates most benefits because it challenges current norms. Benchmarks provide managers with comparative data on performance and labor productivity. Although like-for-like comparisons are not always easy, benchmark measures can give the implementing agency crude indicators of the scale of any overstaffing.

There are three main types of benchmarks:

• Internal benchmarks–By making comparisons within an organization, perhaps between different offices or time periods, it may be possible to identify some areas for improvement quickly and easily. An example is the approach adopted by Kenya's electricity distribution company (see box 3.7).
• Sector benchmarks–Comparisons in the same sector provide another comparison. International or regional comparisons can be used where the PPI enterprise is a monopoly provider in the country.
• Functional (process) benchmarks–There may be other organizations from different sectors but with similar operational functions that can be compared. For example, gas, water, and power utilities might cooperate in benchmarking their metering or billing collection procedures; airlines and railways are similar in the ways they manage the turnaround and dispatch of aircraft or trains; administrative processes, customer service response times, and staff appraisal performance will have similarities in all organizations.

All three types of benchmarks have their places, but a combination of measurement and process analysis is important for effective benchmarking. Measurement identifies the performance gap, but the discussion, debate, and working through of process and operating changes provide the mechanism for operational managers to identify change–including identification of the extent, location, and causes of overstaffing.

To understand the origins of labor productivity, implementing agencies will want to review a range of generic benchmarks (box 3.8), as well as those specific to the sector (box 3.9), such as:

• Number of employees per thousand connections (telephones or water)
• Number of employees per generated megawatt (MW) (for power generation)
• Number of employees per ton of freight or TEU (20-foot equivalent unit) of containers handled (ports)

Sources of Benchmark Data

Benchmarking data can be obtained from international, regional, and national sources. International organizations are one source of benchmarking data, and increasingly make information available for online access through the Internet (table 3.2).

There are also growing networks of collaborating enterprises in the infrastructure and utilities sectors at the regional and national levels. For example, in the water sector:

Link to WUP Web site
Link to Baltics benchmarking Web site

• The WUP provides a regional forum for urban utilities in Africa to share performance data (see http://www.wupafrica.org/). The WUP has just completed a benchmarking exercise gathering data from more than 100 water utilities in Africa.
• In the Baltics a group of utilities are benchmarking against each other (see http://www.water.hut.fi/bench/)
• In Brazil a national agency concerned with water sector reforms, Projeto de Modernizaçao do Setor Saneamento within the federal Ministry of Planning and Budgeting, has a data set on operating costs for about 100 municipalities.
• With World Bank support the Vietnam Water and Sewage Association is creating a database of urban water sector costs and performance in the country through the low-cost collection and publication of data provided by more than 60 provincial water companies (see Nguyen 2002 and benchmarking data for the 60 companies).

Link to AFUR Web site
Link to SAFIR Web site

Other cross-sector regulatory groups, such as the African Forum for Utility Regulation (AFUR), are also trying to include benchmarks and indicators as part of their information-sharing processes (see http://www.worldbank.org/afur/). SAFIR (the South Asia Forum for Infrastructure Regulation) also provides some comparative information (see http://www.safir.teri.res.in/).

In addition to data from international organizations, trade associations, regulators or associations of regulators (such as AFUR or SAFIR), or other groups, statistics on labor productivity may be available from private sector benchmarking firms. Because there is competition among private sector firms in infrastructure services, not all PPI enterprises may be willing to share their methods and commercial information. Private sector intermediaries can provide services in benchmarking and interfirm comparisons, often on a cost-share basis and usually with mechanisms for providing a measure of confidentiality.

Table 3.2: Some Sources of International Benchmarking Information

Sector	Source
Ports	United Nations Conference on Trade and Development (UNCTAD): UNCTAD's Annual Review of Maritime Transport provides statistical data for the world's ports (see the digital library at UNCTAD's Web site, http://www.unctad.org/). Others: Australia's Productivity Commission undertook an international benchmarking of productivity on Australia's ports. Results are available at the Commission's Web site, http://www.pc.gov.au/research/benchmrk/wtfrnt/wtfrnt.pdf. The American Association of Port Authorities (whose members are 150 ports in North and South America and the Caribbean), distributes port statistics and information on labor– management relations. Information is available at http://www.aapa-ports.org/.
Postal operations	Universal Postal Union: The union provides an online database with statistics on variables, including the number of full- and part-time staff. The database is available at http://www.upu.int/. International Postal Corporation: This association of 22 postal operators handling 65 percent of the world's mail undertakes some cooperative benchmarking projects, but to date none of the projects focus on labor issues. The Web site is http://www.ipc.be/.
Rail	World Bank: A principal source of comparative data on worldwide railway performance can be found at http://www.worldbank.org/transport/rail/rdb/countries.htm.
Road	SAFIR: Comparative analysis of bus operations in South Asia (SAFIR 2002) is available at http://www.safir.teri.res.in.
Telecommunications	The International Telecommunications Union (ITU) (http://www.itu.int/): The ITU's statistics department collects aggregate data provided by national ministries or regulators at the country level on numbers of employees in the telecommunications sector. The ITU does not, however, hold data at operator level (although there is a database to facilitate contact with individual operators), and its statistics combine both mobile and fixed line employment. The human resources department of the ITU is establishing regional centers of excellence for training and staff development purposes, and has developed a computerized tool (MANPLAN) for forecasting strategic staffing and training needs. Regional comparative data are available in the ITU's Africa, Asia-Pacific and Americas Telecommunications Indicators reports. The most recent telecommunications indicators from the ITU's statistical database are available from the ITI web site: http://www.itu.int/ITU-D/ict/publications/.
Water	Asian Development Bank: The ADB has financed two issues of the Water Utilities Data Book, which provides valuable information on water utilities in the Asian and Pacific region. The second edition was published in 1997 (McIntosh and Yinguez) and is available for purchase from the ADB, Manila. ADB has provided support for benchmarking in particular regions (e.g., the Pacific Water Benchmarking Study [Delana 2002]). World Bank: The Benchmarking Water and Sanitation Utilities Project has a Web site that provides core cost and performance data: project information can be found on http://www.worldbank.org/watsan/topics/bench/wup.html. Although there is considerable benchmarking activity at the national level, much of the information is scattered. Information on an initiative to help utilities (and regulators) share and access data can be found at http://www.worldbank.org/watsan/pdf/benchmarking/pdf. Included there is a start-up kit for water utilities wishing to participate in benchmarking. A set of water and wastewater utility indicators is available at the Web site of the Water and Sanitation Program, http://www.worldbank.org.watsan/pdf/indicators.pdf. The annual World Bank Water Forum provides a discussion and examples of the use of performance benchmarking, as does the World Bank publication A Water Scorecard (Tynan and Kingdom 2002). International Water Association (IWA): The IWA (http://www.iwahq.org.uk), a forum for sharing of experience among members, recently has published guidelines titled "Performance Indicators for the Water Industry" (Alegre 2000) and "Process Benchmarking in the Water Sector" (Parena, Smeets, and Troquet 2002). The IWA Foundation focuses on water issues in developing countries.

Using Benchmark Measures

The collection and analysis of relevant data (metrics) are essential for the identification of areas of good or poor performance, and for the subsequent analyses of operational processes. This section gives some illustrations of labor productivity benchmarks reported in a number of infrastructure sectors, and makes suggestions on the collection and use of benchmarking data.

Comparisons within a sector can indicate potential low labor productivity, as the following examples illustrate.

Airlines sector:

• Before it was liquidated in the early 1990s the state-owned Zambia Airways employed 300 staff per plane, compared with an industry norm of 140 at that time (Kikeri 1998).
• Loss-making long-haul carrier Air India had a staff-to-aircraft ratio of 663 in 1997, compared with ratios between 170 and 340 in various Southeast Asian carriers: Singapore Airlines, Thai Airways, Malaysian Airlines, and Cathay Pacific (India, Disinvestment Commission 1998).
• At Middle East Airlines (MEA) pilots work 60 hours per month, compared with an average of 90 hours per month in Organisation for Economic Co-operation and Development (OECD) countries. The maximum number of flight hours at MEA is 9 in a 24-hour period, whereas the international average is 10.5 and at some airlines it reaches 12.

Bus operations sector:

• Comparisons of the performance of state bus companies in India showed big differences in labor productivity among states: staff to bus ratios varied from 6.03 in Karnataka and 7.12 in Andhra Pradesh to 11.55 in Madhya Pradesh and 16.08 in Orissa.
• In Sri Lanka private sector bus companies operate with 2 to 3 staff per bus, compared with 5 to 13 staff for state-owned bus companies (SAFIR 2002).

Rail sector:

• A World Bank study of concessioned railways in seven countries showed wide variations in labor productivity, measured as outputs divided by number of employees in the rail sector. The most productive U.S. railways had 22 times fewer workers per traffic unit (passenger kilometers plus freight kilometers) than did some railways in Mexico (Thompson, Budin, and Estache 2001).

  Thompson, Budin, and Estache 2001.

• Following railway privatization, PPI and restructuring the ratio of labor cost to revenues fell from an average of 64.3 percent to 48.4 percent in eight case studies (mainly from industrialized nations) (Kopicki and Thompson 1995).

Telecommunications sector:

• In Malawi the national telecommunications operator has a ratio of 16 working lines per employee compared with a Sub-Saharan average of 31 working lines (Sauti-Phiri 2002).

  Selected telecommunications indicators

Water sector:

• An analysis of data from 246 water utilities (including 123 utilities from 44 developing countries) proposed a benchmarking target of 5 or fewer staff per 1,000 connections for developing-country water utilities. This target was based on the levels of productivity actually being achieved by the top quartile of developing-country utilities within the database. By contrast many developing-country utilities reported more than 20 staff per 1,000 connections (Tynan and Kingdom 2002).
• Comparisons among Vietnam's provincial water companies show a number of operators with labor productivity well below the average (see figure 3.1), which would justify further assessment of the cause.

Making comparisons within a region can also prove valuable for implementing agencies that need to understand whether overstaffing is confined to just one enterprise or is a common problem in all infrastructure utilities. As shown in table 3.3, an assessment of the utility sector in Uruguay compared with other countries in the region signaled potential problems in labor productivity in a number of utility sectors. Regarding that table, two points are instructive:

1. Differences in productivity within a geographic region can be substantial. Implementing agencies do not need to compare between OECD industrial countries and developing countries to gain useful insights. The benchmarks and the comparators in this example were all classed in the 1997 World Bank World Development Report as upper middle income countries (except the Republic of Korea, which then was classed as lower middle income)
2. There is some degree of subjectivity. The basis for assessing the "best performance" benchmark was based on a range of sector and regional reports plus interviews with sector specialists for each country.

Table 3.3: Regional Comparative Performance Measures

Sector	Uruguay	Argentina (private sector)	Brazil	Mexico	Chile	"Best performance" (and reasonable benchmark)
Telecommunications (main lines per employee)	88	155	121	174	153	294 (Korea, Rep. of)
Electricity (customers per employee)	102 (in 1995)	280	177	208	285	285 (Chile)
Water and sanitation (employees per 1,000 connections)	8	3	5.2	4	2.1	1.8 (Malaysia)

Source: World Bank 1997.

In practice, several factors make the comparison of benchmarks across countries and PPI operations challenging. These factors include:

• Increased outsourcing and contracting out. Because utility and infrastructure enterprises outsource many of their operations, comparisons based on units of activity per fulltime, permanent employee may not provide a like-for-like comparison.
• Comparability of the scope of the PPI enterprise. Published data may report labor numbers and productivity in operations that are combined in some countries and separated in others–for example, telecommunications and postal services; water and sewerage operations; and power generation, transmission, and distribution.

  Labor productivity "norms" may change quickly, especially following the introduction of competition or rapid growth in demand for services.

• Differences in condition of the infrastructure. Some older networks have high maintenance costs as a result of age or past inadequacies in investment in new technologies (be it optical fiber for telecommunications, port containers, or combined-cycle power plants).
• Extent and nature of the network. Service providers in dense urban areas will have staffing requirements that differ from those of rural providers. Some railways may have a markedly more benevolent topography than others, so that track maintenance costs are lower. Different regulatory regimes may place different legal obligations on the level of service provision, leading to very different cost and staffing structures.
• Depth and quality of the data. All benchmarking data sets will benefit from greater precision, clear definitions, and disaggregation. The more information that is available and the more that users can be sure of the relevance of the data sets, the more trust can be placed in them. Even so, averages can be deceptive and can be distorted by abnormally high or low performance.

  Benchmarks change constantly as technologies and work practices change.

• Misuse and abuse of benchmarks. Labor benchmarking statistics can be misused and used to exaggerate or understate the need for downsizing. For example, simply setting labor adjustment targets to match international best practice levels can be dangerous if it does not take account of the particular conditions bearing on the enterprise. Furthermore, data can be manipulated (for example, by excluding temporary or seconded workers) to suggest that staffing levels are not particularly high.
• Age of the data and the fast-changing nature of the work force. Almost by definition, a benchmark will be out of date the day it is published. One year's best practice can soon translate into next year's average performance so it is essential to ascertain the date relevancy of the data. Old data are still valuable, however, because they allow trends to be identified, thus enabling the implementing agency to assess whether productivity and efficiency gains are accelerating or stagnating. One example of changing productivity is that of Bharat Sanchar Nigam Limited (BSNL)–the main state-owned telecommunications operator in India. As the number of subscribers has risen, staff numbers have remained constant and labor productivity has risen steadily (table 3.4).
• Policy and structural reforms in the sector. Productivity benchmarks also change as a result of liberalization, new entrants into the sector, and new technologies. For example, a private sector operator and new entrant into telecommunications in India, Tata Teleservices, has about seven employees per 1,000 subscribers for the fixed services it provides in Andhra Pradesh (albeit using radio for the local loop)–approximately half the ratio achieved in 2000 by the former monopoly, BSNL. In general, as infrastructure companies are exposed to competition and new investment is increasing, the work force in benchmark comparators changes from year to year as a result of increased demand for very experienced managers and senior specialists with commercial, financial, and information technology skills; fewer unskilled workers but more workers with technical skills and experience in newer technologies (especially in sectors such as telecommunications); and fewer administrative and clerical jobs, but more customer service facilities.

Table 3.4: India–Changing Labor Productivity at BSNL, Selected Years

Year	Subscribers (millions)	Employees (thousands)	Employees per 1,000 subscribers	Subscribers per employee
1981	2.15	289	134.45	7.44
1991	5.07	375	73.90	13.52
1995	9.80	419	42.78	23.39
1996	11.98	421	35.14	28.46
1997	14.54	429	29.50	33.89
1998	17.80	425	23.87	41.88
1999	21.59	424	19.64	50.92
2000	26.51	421	15.88	62.97

Source: Reports issued by BSNL.

In summary, the key to choosing and using benchmarks for labor adjustment is in selecting operations and measures that are as comparable as possible. The development of regional, national, and international benchmarking and information-sharing groups is likely to improve the availability, quality, and relevance of data. Comparative benchmarking provides valuable information on potential levels of overstaffing, even if it is best used in combination with other analyses. (See box 3.10 for a list of suggestions for making the best use of benchmarking data.)

Benchmarking Labor Costs

Given the difficulties in comparing labor productivity in terms of output per employee, one alternative approach is to focus more on benchmarks involving output per unit cost of labor or labor costs as a proportion of total operating costs.

In the rail sector, even comparisons of partial labor productivity measures are difficult because of differences in topography, traffic mix, technology, level of past investment, international trade disruptions, industrial geography, and so on. Basic measures such as ton-kilometers, passenger-kilometers, locomotive-kilometers, revenue ton–kilometers are, more often than not, estimates based on tons of freight or passengers multiplied by average length of haul or trip. The difficulty in calculating passenger- kilometer estimates is particularly great on railways with many urban commuters. Combined measures such as traffic units per employee (tonkilometer+ passenger-kilometer) suffer from similar problems.

Examining staff costs (wages plus benefits) as a percentage of total operating revenue reveals that, in a number of railways, staff costs alone exceed total revenues from users and are often the largest single cost category. This may be a better way to evaluate labor productivity than using a ratio of staff to traffic units (passenger kilometers plus freight kilometers) because it factors in differences between labor unit costs in different countries, which might be a reason for some railways legitimately being more labor intensive than others.

In some sectors labor costs are relatively low as a proportion of operating costs or capital costs. In an analysis of 77 electricity-generating plants in 28 industrial and developing countries, the average shares of cost were 10 percent for lubricating oil and materials and 13 percent for labor, but 48 percent for fuel and 29 percent for capital (Diewert and Nakamura 1999, based on a 1993 data set.)

Figure 3.1 provides a simplified performance structure for a generic utility and shows that labor costs are only one part of the overall cost structure.

In a few cases the PPI investor may not be very concerned about staff numbers, perhaps because there is little overstaffing or because low wages make labor a small proportion of overall costs.

More commonly, however, overstaffing means low labor productivity and high staffing costs. Low wages do not necessarily mean relatively low staff costs. In their analysis of 246 water utilities, Tynan and Kingdom (2002) found large differences within developing countries. Average staff costs as a proportion of total operating costs were 39 percent in developing-country utilities compared with 29 percent in industrialized-country utilities.

Even where labor productivity is poor, other factors play a part. For example, high water tariffs in Conakry, Guinea, were only partly the result of low labor productivity (by regional standards). High debt-servicing costs, considerable amounts of bad debt, low collection rates, and a high percentage of expatriate staff were other factors (Brook and Smith 2001). (That is why some governments and firms prefer to measure "total factor productivity" as a more accurate guide than raw output/input ratios on numbers or costs of workers; see, for example, Cowie and Riddington 1996, and Economic Commission for Europe 2002).

Home

How to Use the Toolkit

Labor Toolkit:
Framework and Overview

Labor Impacts of PPI

Assessing the Scope of Restructuring

Overview

Staff Audits

Benchmarking

Work Force Analysis

Who Should do the Assessments?

Material and Sources

Strategies and Options

Key Elements of a Labor Program

Engaging with Stakeholders

Monitoring and Evaluation

Sitemap

Search   

Download Modules as PDF Documents

References

Glossary

Case Studies