Challenges in the Measurement of Public Sector Productivity in OECD Countries

Transcription

1 Challenges in the Measurement of Public Sector Prouctivity in OECD Countries Ewin Lau, Zsuzsanna Lonti an Rebecca Schultz OECD 1 ABSTRACT Prouctivity is one of the main engines of economic growth. While most existing work on prouctivity has focuse on the private sector, there is great value in better unerstaning prouctivity in the public sector, given government's important role in the provision of goos an services an its substantial contribution to overall GDP. However, the measurement of public sector prouctivity as a first step towars better unerstaning its ynamics is fraught with challenges, as the public sector iffers substantially from the private sector in some of its key characteristics. This article examines current country practices an challenges to measure public sector prouctivity an ientifies five areas to further enhance measurement efforts: (i) improvements to input measurement an cost accounting, (ii) stanarization an comparability of measures, (iii) output measurement beyon the eucation an health sectors, (iv) a typology of activities at the micro level, an (v) intra-governmental co-orination on prouctivity measurement. The article also calls for further research on the policy rivers for public sector prouctivity to elve eeper into how governance frameworks can be mobilize to achieve greater public sector prouctivity in support of effective public governance an ultimately the well-being of citizens. Prouctivity, unerstoo as the volume of output prouce for each unit of input, is one of the main engines of economic growth. While most OECD work on prouctivity evelopments has focuse on the private sector, there is great value in better unerstaning prouctivity in the public sector, given that government is a main provier of goos an services to citizens an government prouction is responsible for a substantial share of GDP. However, unerstaning public sector prouctivity poses several challenges. The public sector iffers substantially from the private sector in some of its key characteristics. This article examines current practices an challenges to measure an improve public sector prouctivity an suggests ways forwar to aress measurement challenges an elve eeper into how governance frameworks an processes can be mobilize to achieve greater 1 Ewin Lau is the Hea of Reform of the Public Sector Division, Public Governance Directorate, OECD; Zsuzsanna Lonti is a Senior Project Manager Government at a Glance, Public Governance Directorate, OECD; an Rebecca Schultz works as a Policy Analyst, Public Governance Directorate, OECD. This article is an abrige an revise version of Lau (2017). This article buils on the work currently unertaken by the OECD s Public Governance Directorate on public sector prouctivity. The authors woul like to thank Luiz e Mello, Alessanro Lupi an Delphine Moretti from the OECD s Public Governance Directorate for their contribution to this article an Peter van e Ven from the Statistics Directorate of the OECD as well as the anonymous reviewers an the eitor of the journal for their useful comments. s: Ewin.LAU@oec.org, Zsuzsanna.LONTI@oec.org, an Rebecca.SCHULTZ@oec.org I N T E R N A T I O N A L P R O D U C T I V I T Y MON I T O R 180

2 public sector prouctivity that supports effective public governance an ultimately the well-being of citizens. Prouctivity refers to how much output is prouce for each unit of input, calculate as the ratio of a volume measure of output to a volume measure of input use (OECD, 2016a). 2 Raising prouctivity the ability of economic actors to prouce more outputs with better-combine inputs, or use fewer inputs to provie the same outputs is the engine of economic growth. Improving public sector prouctivity is high on many countries' political agenas. 3 OECD countries are facing significant emographic challenges with their aging populations an increasing epenency ratios that will affect both the eman for public services an the capacity to eliver them. In orer to accommoate these evelopments, either more resources are neee or the prouctivity of the public sector has to increase. Also, most OECD countries are still experiencing fragile public finances with high ebts an continuing eficits, so there is still a nee for retrenchment over the meium term. Public managers are oblige to maximize the return to the public, incluing making the most out of the available talent in the public sector, an are accountable to the citizens for the efficient operations an results achieve of the public sector. UK estimates show that ifferent rates of growth in public sector prouctivity have significant effects on the public sector ebt relative to GDP (DCLG, 2015). Moreover, trust in government is also eclining in OECD countries, especially in those countries that were the harest hit by the last economic an financial crisis. The provision of better quality public services through better resource utilization which means increasing prouctivity coul help improve citizens' views on government, an more specifically on the public institutions proviing these services. Public sector prouctivity has a significant impact on the performance of the national economy an societal well-being. First an foremost, the public sector is a major irect proucer of goos an services: on average government prouction costs represents 21.9 per cent of GDP across OECD countries an gross value ae of government amounte to 12.3 per cent of GDP in 2015 (Chart 1). 4 Governments are the main, an sometimes only, proviers of key goos an services, such as eucation, health, social services, transportation an infrastructure. In fact, OECD governments are responsible for 70 per cent of final consumption expenitures 5 on health goos an services an for 84 per cent of final consumption expenitures on eucation. In aition, the public sector is a key enabler of the 2 As such, the concept of prouctivity is istinct from relate concepts like "value for money", which "implies that reforms must lea to better quality of services for citizens an businesses or to savings, or to both" (OECD, 2015c), or cost-effectiveness, which refers to the extent to which an activity attains its esire objectives (OECD, 2013). 3 Public sector in this analysis refers to "general government" as efine in the System of National Accounts, which oes not inclue public corporations that prouce primarily for the market. It encompasses central government, state an regional government, local government an social security funs. "Government units are unique kins of legal entities establishe by political processes that have legislative, juicial or executive authority over other institutional units within a given area. Viewe as institutional units, the principal functions of government are to assume responsibility for the provision of goos an services to the community or to iniviual househols an to finance their provision out of taxation or other incomes, to reistribute income an wealth by means of transfers, an to engage in non-market prouction (EC, IMF, OECD, UN an Worl Bank 2009:4:17). 4 Prouction costs encompass compensation of employees, goos an services use an finance by government (e.g. outsourcing) an other prouction costs, which inclue the consumption of fixe capital (epreciation of capital) an other taxes on prouction less other subsiies on prouction. Gross value ae is the ifference between gross output an intermeiate consumption. 181 N U M B E R 32, S P R I N G 2017

3 Chart 1: Government Prouction Costs an Gross Value Ae as Percentage of GDP, 2015 Prouction costs Gross value ae Source: OECD National Accounts Statistics (atabase). Data for Australia are base on a combination of Government finance statistics an OECD National Accounts ata provie by the Australian Bureau of Statistics. Note: The full names of the non-oecd countries are: ZAF: South Africa, RUS: Russian Feeration, LTU: Lithuania, COL: Colombia, CRI: Costa Rica. Please note that the country coes use are official ISO coes, which are available at: Both numerator an enominator of Chart 1 are in nominal terms. Thus, the GDP is in nominal terms. The OECD average is calculate as a weighte average where the weight is represente by the enominator of the ratio (i.e. the GDP of the countries). proper functioning of the economy an society. For example, investment in eucation an in infrastructure both impact the prouctivity of the economy as a whole. A well-performing public sector also contributes to higher overall prouctivity through goo quality regulation, the absence of corruption, an soun public financial management. Public sector prouctivity cannot be unerstoo without the ability to measure it, which requires goo quality an, if possible, internationally comparable input an output measures of public sector services. Consierable progress has been mae in the last two ecaes in the measurement of public sector inputs an outputs in the framework of the System of National Accounts by National Statistical Offices (NSOs), governments an their epartments, national Prouctivity Commissions, international organizations, such as Eurostat or the OECD. But there is still much to be one. To ate, the lack of measures to appropriately capture public sector prouctivity builing on, an going beyon, the System of National Accounts, has meant that major policy ecisions are being taken without aequate unerstaning of their implications for the economy as a whole. Too often, improving public sector prouctivity is equate simply with spening or staff cuts. The term prouctivity is often misuse as a syn- 5 Final consumption expeniture represents the amount spent by governments, non-profit institutions an househols on goos an services consume. The corporate sector oes not incur any final consumption expeniture because it only prouces final goos for sale in the market. Expenitures are attribute to the institutional unit (government, non-profit institution or househol) that bears the costs (note that nonprofit institutions represent a very small portion of the total consumption). Compare to total expenitures, final consumption expenitures exclue spening on goos an services not consume uring the year, such as investment goos, an exclue social benefits provie to househols which are not tie to the consumption of specific goos an services, such as pension payments. I N T E R N A T I O N A L P R O D U C T I V I T Y MON I T O R 182

4 onym for austerity program, rather than searching for strategic agility, improving the mix an use of inputs, an enhancing the quality of outputs for better public outcomes. A better uners t a n i n g o f p u bl i c s e c t o r p r o u ct i v i t y measurement challenges can therefore provie insights into how public sector prouctivity can be improve. This article is base on a short survey carrie out by the OECD at the en of 2016 an early 2017 on country practices in prouctivity measurement. 6 The survey was esigne as a mapping exercise to collect basic information on countries' general efforts to measure public sector prouctivity from a managerial perspective. Data presente are self-reporte an aim at capturing countries' (exploratory) managerial efforts to measure public sector prouctivity rather than measuring public sector prouctivity in their national accounts systems. The survey was sent to senior buget officials an prouctivity commissioners in OECD member an partner countries to collect information on (i) the methoologies applie to measure public service prouctivity, (e.g. use of cost accounting an quality ajustment); (ii) targete sectors, organizations an functions; (iii) main organisations responsible for this work; an (iv) the use of prouctivity measures in government. Thirty OECD member an two partner countries respone to the survey. The article first presents an overview of existing efforts to measure public sector prouctivity, followe by an analysis of the specificities of prouctivity measures for the public sector. It conclues with five concrete suggestions to enhance existing measures of public sector prouctivity an reflects on the role of measurement to support further research on ways to improve public sector prouctivity. Which Aspects of Public Sector Prouctivity Be Measure? Prouctivity in the public sector can be measure at several levels epening on the focus of the inquiry. At the macro level, the prouctivity of the whole public sector is calculate, which allows for the estimation of the public sector's contribution to the performance of the whole economy an to a more accurate estimate of GDP growth. At the meso level, the focus is on the prouctivity of the various policy sectors, such as eucation or health. At the micro level, the performance of iniviual organizations e.g. ministries, agencies, hospitals, schools, municipalities as well as key activities an functions, e.g. share services, procurement or waste collection etc. can be compare not only within countries but also across countries. Measurement on all three levels face similar methoological challenges, such as the ientification of the concrete services provie, separating out the inputs expene on those services an effective ways to capture changes in quality over time, etc. A recent OECD survey has collecte information on countries' current efforts to measure public sector prouctivity on the ifferent levels. 7 Macro level Macro level measurement of prouctivity is useful to establish trens over time in one country, an also compare those trens across countries. It is less useful for management purposes, as macro-level ata encompasses a iverse set of activities of numerous units. Given its aggregate 6 The artcile also raws on a review of recent evelopments in the measurement of public sector prouctivity commissione by the OECD (Robano, 2016) an a paper prepare for the Meeting of the Performance an Results Working Party (Dunleavy, 2016). 7 See Box 1 in the online Appenix available at: N U M B E R 32, S P R I N G 2017

5 Chart 2: Measurement of Public Sector Prouctivity at Different Levels ' E e e e e e e ^, ^ : K e e /, D > ^ D K e W, W K Source: 2016 OECD Survey on Measures of Prouctivity in OECD an Partner Countries nature, it also masks countervailing evelopments in ifferent parts of the public sector an, as a result, it is not actionable as it cannot clearly be attribute to a particular part of the public sector. The prouctivity of the whole public sector 8 is measure only in a few OECD countries accoring to the OECD survey. 9 In 2015, only seven countries (Australia, Denmark, the Netherlans, New Zealan, Portugal, South Africa an the Unite Kingom) reporte measures of the prouctivity of the whole public sector (Chart 2). The UK release the first total public sector prouctivity estimate in 2009 (for current measurement efforts, see Box 1) an Denmark starte its measurement in 2014 (for etails, see Box 1). Finlan has stoppe measuring prouctivity of the whole public sector an concent r a t e s o n m e a s u r i n g i t o n l y f o r t h e municipalities, where the greatest prouctivity gains can be realize. Meso level At this level, most efforts to ientify irect output measures by countries an by international organizations have focuse on two policy sectors that provie iniviual services: eucation an health. 10 Direct output inicators allow the measurement of prouctivity of these services. These inicators facilitate the unerstaning of the contribution of these policy sectors to the prouctivity of the public sector 8 Please note that the OECD mapping survey oes not istinguish between iniviual an collective services when referring to the public sector. 9 At the same time, many OECD countries inicate that they have some form of measurement of the outputs of government or specific sectors of government. This is ue in part to the Eurostat requirement for EU member countries to measure an report the outputs of their non-market services. Although such measures o not represent full-flege prouctivity measures that relate outputs to the inputs use to generate them, this is an important first step towars measurement of public sector prouctivity. 10 See Box 1, Table 1 an 2 in the online Appenix available at: Lau%20Appenix.pf I N T E R N A T I O N A L P R O D U C T I V I T Y MON I T O R 184

6 Box 1: Report on Public Sector Efficiency - Unite Kingom The UK government carrie out its first comprehensive review of public sector efficiency in Base on UK an international experience, the review gathere existing evience on efficiency trens an rivers an ientifie areas for achieving further efficiency improvements. The report efines public sector prouctivity as a key component of public sector efficiency. Public sector efficiency refers to the "entire process of turning public money into esire outcomes." It can be ivie into "technical efficiency", which focuses on "oing things right", e.g. by oing things at lower costs or proucing more outputs from what is currently one at the same cost; an "allocative efficiency", concentrating on "oing the right things", i.e. fining ways of achieving esire outcomes at less cost. Public sector prouctivity is efine as the relation of "how much output is prouce for each unit of input", an hence forms part of technical efficiency. The analysis shows that public sector prouctivity growth is essential for elivering fiscal consoliation. Although market sector prouctivity growth has outstrippe prouctivity growth in the public sector between 1997 an 2010, public sector prouctivity has grown by 3.7 per cent since A one per cent increase in annual public sector prouctivity woul imply a reuction in public sector net ebt of 64 per cent of GDP by the first quarter of The report presents prouctivity trens in all areas of public sector activity, incluing a wie range of frontline services. Output measures are quality-ajuste for some health an eucation services. The analysis reveale a number of problems an challenges in measuring public sector efficiency. These inclue a lack of comprehensive an comparable evience, the efinition an measurement of output quality, an the attribution of changes in outcomes to changes in outputs. Accoring to the report, sharing evience, examples an best practice across government has prove useful for informing action to further improve efficiency. The report also recommens the use of triangulation, i.e. to not rely on a single measure, but to raw on a range of evience when analyzing public sector efficiency. As a result, the UK government has prouce an Efficiency Toolkit proviing guiance on improving efficiency in the public sector, an a number of practical steps to isseminate the report's finings an inuce follow-up conversations on the topic are suggeste. Source : Department for Communities an Local Government (DCLG) (2015); Robano (2016) an the whole economy. Furthermore, countries' performances can be benchmarke through international comparisons. In 2015, 11 countries reporte measuring the prouctivity of eucation, an 12 countries measuring the prouctivity of health care services. Among the countries that reporte measuring public sector prouctivity, only Israel, Polan an Portugal report not having unertaken measurement efforts in either of these sectors accoring to the OECD survey. With the sprea of output measurement in these sectors in OECD countries an avances in price measurement an further successful stanarization efforts, the eucation an health sectors are ripe for prouctivity measurement in a stanarize way. Micro level Governments are also intereste in how well i.e. how effectively an efficiently their organizations work an carry out their key functions. Prouctivity measurement at the microlevel is most useful for managerial purposes, as managers shoul be accountable only for the performance of units that they control. It is also 185 N U M B E R 32, S P R I N G 2017

7 Box 2: Measurement of Government Outputs an Prouctivity in Denmark The Government of Denmark in 2014 move from the use of an input-base metho to an outputbase metho for measuring the volume of general government prouction. A total of 18 volume inicators have been constructe to measure iniviual non-market services in the areas of health care, social protection, eucation, an recreation an culture. Only for collective services (e.g. efence, public orer an safety, environmental protection), which account for about a fourth of general government output, is the input-base metho still use. This implies that for collective services, any potential changes in prouctivity levels are not capture, as the output volume is measure by the volume of inputs. The volumes of non-market services calculate base on the input- an the output-base metho clearly iffer from each other. Between 2008 an 2014, the volume growth is higher in five out of the seven years using the output metho. This may suggest an increase in prouctivity, as the volume of services provie by the general government (output) increases more quickly than the volume of resources it consumes (input). However, the Government of Denmark points out that this conclusion may be misleaing in the short run, as phenomena like changes in legislation may lea to rapi changes in output levels implying prouctivity increases (e.g. new legislation resulting in greater numbers of stuents) that are not matche immeiately by the corresponing necessary inputresources (e.g. builings, teachers). Results base on the output-base metho show that general government contributes significantly to general economic activity in Denmark. General government prouction represents about 15 per cent of the whole economy. Labour prouctivity, measure as the gross value ae per hour worke without quality ajustments, increase by 1.0 per cent in general government over the perio, while it increase in the market economy by 5.2 per cent. Statistics Denmark worke in close co-operation with the Ministry of Finance an the responsible line ministries, such as the Ministry of Eucation an the Ministry of Health, to select, evelop an valiate the output inicators. Acaemic experts were also involve in the elaboration of the methoology. Eurostat is another important partner for Denmark in their work on measuring government output. Participation in Eurostat task forces on the evelopment of measurement methoologies have been an important input for Statistics Denmark's approach to measuring the volume of government services. In line with EU requirements, the Government of Denmark is currently not applying quality ajustments to the volume measures use. However, Denmark is involve in efforts to scope methos for quality ajustments in the future through a working group in Eurostat as well as within Statistics Denmark. Source: Statistics Denmark (2013), "General Government Output an Prouctivity " en/statistik/publikationer/vispub?ci= useful for making comparisons over time an for target setting within an organization. It allows comparing their performance to each other within a government an to similar organizations in other governments. Certain functions such as tax or pension aministration or foreign policy are carrie out only by a single organization in each country, so bench- I N T E R N A T I O N A L P R O D U C T I V I T Y MON I T O R 186

8 marking them can only be one internationally. Consierable work has been one for example on the international comparison of tax aministrations since 2006 in OECD countries (OECD, 2015b). However, while the manate of these tax aministrations might be similar, their functioning an prouctivity is largely epenent on the tax policies an the tax coes that they aminister. As a result, international comparisons of their efficiency/prouctivity coul pinpoint changes that might be necessary not only in tax aministration, but also in tax policy. In the same vein, the prouctivity of a hospital is strongly epenent on its case an function mix, the socio-economic backgroun of its patients an other factors, an not just on how efficiently it carries out isease treatments. This is why comparisons of performance on isease treatments across hospitals shoul be interprete with caution. However, the collection of prouctivity ata for schools, hospitals or municipalities coul also prouce greater insights into the possible causes for their ifferences in performance. With the increase of the number of observations, more sophisticate econometric techniques can be applie, such as fixe effects regression analysis an stochastic frontier analysis (Dunleavy, 2016). Governments may also want to compare how prouctive certain functions/practices are. This type of measurement is in its infancy. The selection of the functions for which analysis shoul be carrie out coul be base on the following principles: the functions are carrie out by most governments; they contribute substantially to the prouctivity of the public sector; inputs an outputs of the function can be efine; aministrative ata on these inputs an outputs are reaily available; an ifferent moels exist on how the function is carrie out. An area that fits these principles is the public procurement function. A pilot project on how to measure the prouctivity/value for money of the public procurement function is planne in a number of OECD countries as part of the OECD work on Civil Service Effectiveness. Opinions on the usefulness of measuring the prouctivity of agencies iverge. Dunleavy (2016) avocates strongly for measuring prouctivity at the agency level an comparing prouctivity paths (trens over time). He consiers transactional services the most promising area for prouctivity measurement an makes the case that the measurement of prouctivity of collective services e.g. efence or police is not insurmountable. He also argues that the long-run inability to evelop wiely use measures of government prouctivity reflects on one han a consierable failure of imagination an focus in economics an public management stuies, an on the other han some very sustaine resistance by civil servants, public sector professionals an some politicians to the application of crue an limite measures of government activities. Specificities of the Public Sector an its Impact on Prouctivity Measurement The measurement of public sector prouctivity is rife with consierable ifficulties that emanate from the specificities of the public sector. First, the public sector provies mainly services such as eucation, health, social services, policing, etc. an the measurement of service output is more complicate than the measurement of the prouction of goos, even in the private sector. Services are also heavily reliant on intangibles an tacit knowlege an are often process-base. Furthermore, a large share of government services are not bought an sol, 187 N U M B E R 32, S P R I N G 2017

9 Chart 3:Government Expenitures on Iniviual an Collective Goos an Services in EU Countries as Percentage of Total Expenitures, % 90% 80% 70% 60% 50% 40% 30% 20% 10% Iniviual goos an services Collective goos an services 0% DNK NOR FRA FIN SWE AUT LUX EST DEU ITA SVN NLD GBR CZE ISL ESP PRT POL IRL HUN GRC Source: OECD National Accounts Statistics (atabase). e.g. they are non-market goos an services an, as a result, market prices o not exist. Finally, a significant part of government services representing close to 30 per cent of public spening are collective goos which cannot be consume iniviually (e.g. efence, police, environmental protection) (Chart 3). Outcomes like societal well-being, economic growth an social inclusion are the ultimate goals of any public policy making. The measurement of policy outcomes is not particularly ifficulties in many cases (e.g. citizens' health status, eucation levels of pupils). However, it is much more challenging to establish a clear link between public policy action an policy outcomes, as outcomes can be influence by a range of other external variables beyon output changes that are ifficult to track. For the time being, the European System of Accounts specifically exclues output measures ajuste for quality base on outcomes, in orer to preserve the international comparability of results (EU, 2016: 38). Input Measurement Input inicators measure the amount of resources use for the prouction of outputs. Major input categories are compensation of employees, use of capital goos, use of intermeiate goos (intermeiate consumption) an taxes net subsiies. The basis of prouctivity measurement for the public sector an the unerlying policy sectors is the ata provie by the System of National Accounts (SNA) on inputs an outputs. Here, it is important to note that both measures nee to be ajuste for price changes to arrive at an appropriate measure of prouctivity change, which refers to the change in the volume of outputs, as compare to the change in the volume of inputs. It shoul also be acknowlege that volume changes shoul reflect both changes in the quantity an changes in the quality of the inputs/outputs uner consieration. In the public sector, measurement of government inputs (Box 3) remains ifficult for three main reasons: In most countries bugets are prepare on a cash basis an full overhea costs (incluing capital consumption an occupancy costs I N T E R N A T I O N A L P R O D U C T I V I T Y MON I T O R 188

10 Box 3: Measurement of Government Inputs For inputs, the major methoological issues relate to the appropriate calculation of volume measures, especially for labour (compensation of employees) an the consumption of fixe capital. The quantity of employee labour is efine as "an hour's work of a given type of level of skill" (EU, 2016). The price of labour inclues all changes in compensation that are not relate to skill-ajuste hours worke. The European Union (2016) suggests two methos for estimating volume changes for labour: (i) irectly by measuring the quantity of hours worke, or (ii) eflating money wages an salaries. While the European Commission recommens using the latter approach, most countries use the quantity of hours worke as a measure of labour volume. The most significant challenge is ajusting the change of the labour volume measures for the ifferent skill levels of the employees. The measurement of capital inputs use by government is not wiesprea among. Schreyer an Mas (2013) note that, while capital epreciation is accounte for when measuring non-market output, no other capital input, such as the returns on capital investe, is compute. Accurate measurement of inputs at any level requires cost accounting, which in turn buils on an accrual-base accounting system that registers costs an not only cash-flows. By 2016, aroun three quarters of OECD countries have aopte full accrual accounting (Moretti, 2016). For costs to be apportione to outputs, it is not only necessary to maintain a system of accrual accounting, but also to have a management accounting system that allocates costs to the ifferent outputs. This requires a financial aministration at the agency level that recors the costs of all outputs prouce, which are also neee for the calculation of output volumes. for public builings) incurre by government entities are not measure. Even for those countries that have aopte accrual bugeting or accounting, some assets (historical builings) an the costs relate to the increase of liabilities regaring future pensions an benefits to public servants may not be recore in the accounts. Accounting rules are not consistent across government. Government entities o not always use the same rules in accounting for their operations; hence there exists little consistency in measuring costs. Cost accounting is little evelope. Cost accounting consists of attributing costs (commonly calle expenitures) to core outputs. Difficulties commonly arise in the public sector because: i) spening is presente in buget ocuments by aministrative units or by nature of expeniture, instea of being allocate to outputs; an ii) significant costs, such as public infrastructure investment, are ifficult to allocate to a specific aministrative unit, an even more so, to a specific output (for example, IT systems). Despite these ifficulties, a number of countries have mae significant progress in measuring inputs an linking them with outputs, either at the micro level (for example, the UK, see Box 2) or at the macro level (for example, France). For example, moern methos to estimate labour input inclue irect measures of hours worke, hours worke by the 'full-time equivalent' number of employees, or the hours pai. Using the total number of employees is not consiere to be a suitable inicator. There are also several methos to ajust labour input measures for the ifferent skills of employees, e.g. skill levels for certain types of work can be classifie by eucational level, occupation, salary or grae (EU, 2016:66). 189 N U M B E R 32, S P R I N G 2017

11 Chart 4: Structure of General Government Expenitures by Function in OECD Countries, 2014 ' e ^ e W e Z, e, Source: OECD National Accounts Statistics (atabase); Eurostat Government finance statistics (atabase). Data for Australia are base on Government finance statistics provie by the Australian Bureau of Statistics. Output Measurement Outputs are the goos an services prouce (e.g. number of hours chilren are taught). For prouctivity calculations final outputs shoul be measure (e.g. the proucts or services that are prouce for the public or business) in orer to avoi ouble counting. However, it is not easy to ientify final outputs ue to the absence of prices. Due to the peculiarities of the public sector, prouctivity was long consiere constant base on the convention that in the public sector inputs equal outputs. With this convention, until not that long ago OECD countries measure the volume (i.e. economic value) of nonmarket services by the input-base metho. This has change in recent years with an increase interest an focus of policy makers on performance of the public sector. Many countries measure activities now that reflect what the non-market units actually o with their inputs (ONS, 2005), but these are imperfect measures. For example, the treatment of appenicitis use to require an operation an a two-week hospital stay. Now, patients only nee to stay in hospital for three ays following the operation. Using an activities-base approach to measuring output, this woul imply a ecrease of output an prouctivity when it is clearly not the case. In aition, the quantities of ifferent outputs nee to be ae together using ata on prices or, in the absence of prices, average unit costs. In current practice, aggregation of outputs is usually base on average costs to facilitate measurement. However, recently it has been suggeste that replacing average costs by the time use to prouce the output coul better reflect marginal utility to consumers (Diewert, forthcoming). Such an approach, however, woul require etaile an accurate information of the time neee to prouce outputs. With avances both in input an output measurement an with the introuction of the output-base metho, 11 some countries (e.g. the Unite Kingom, Finlan, New Zealan, an Australia) starte measuring public sector prouctivity estimates. This is especially the case in some key policy sectors like eucation an health care, which provie services for iniviuals where outputs can be ientifie. However, I N T E R N A T I O N A L P R O D U C T I V I T Y MON I T O R 190

12 for collective services, the ientification of outputs is extremely complicate, an the use of input metho is still common practice (EU, 2016). Similarly, a lack of stanarize methoologies presents a challenge to international comparisons of public sector prouctivity at one point in time. For now, only evelopments in public sector prouctivity over time within countries can be compare. The Atkinson Review (ONS, 2005) was unertaken in the UK an pioneere important methoological evelopments that constitute a significant avance in the measurement of public sector outputs. Since 2006 Eurostat has been requiring EU members to measure an report the outputs of their non-market services, a prerequisite of prouctivity measurement. The OECD has also been working on proviing avice on how to measure the volume of output for eucation an health services (OECD, 2010) in a first attempt to internationally stanarize output measures for these two sectors, an to evelop output price inexes for some services (e.g. health). Government's spening on health care an eucation in OECD countries represente on average 18.1 per cent an 12.5 per cent of total government expeniture respectively in 2014 (Chart 4). These services also are very important for economic growth as well as for iniviual well-being. The ata require to ientify an measure the outputs are commonly base on aministrative sources, for the most part reaily available, but in some countries also complemente by survey ata. Direct output volume measurement requires information on prices, or in the absence of prices unit costs average costs per unit of output 12 an quantities of homogeneous proucts. The measurement of volumes has three imensions: it nees to take account of variations in quantities, variations in the composition of the aggregate prouct, an variations in the quality of goos an services (EU, 2016). There are ifferent practices followe by OECD countries to apply quality ajustment. 13 Output measurement is critical to prouctivity measurement. While outputs refer to goos an services prouce by the public sector, from a performance perspective, what really matters are the outcomes, i.e. the results that policy makers inten to achieve through the prouction of those public goos an services. Countries often have important ifferences in their efinition of outputs an/or outcomes for the same activities. For example, in the case of eucation, in some countries the output is ientifie as the number of pupils; in others, the number of hours chilren are taught. But chilren go to school to learn. As a result, the outcome for eucation is consiere to be the performance of stuents, in practice measure by stanarize tests, such as PISA. While the quality ajustment of outputs may seek to take achievement of outcomes into account in the measurement of outputs, in practice, it is often har to elineate outputs an outcomes an even harer to ientify outcomes that can be solely attribute to public sector activities an their outputs. For example, the achievement of stuents on stanarize tests are not solely the results of the number of hours they are 11 In the output-base metho units of quantity for non-market services are efine. By oing this, it is possible to apply the general principles for calculating volume inices for those services. This can be one in two ways: irectly by calculating a weighte average of quantity changes in a homogeneous number of representative outputs/activities (i.e. irect volume) or by eflation, using as eflator of a price inex an unit cost inex (i.e. inirect volume). 12 The average costs per unit of output oes not inclue changes in the prices of intermeiate proucts (Lorenzoni, 2015). 13 See Box 2 in the online Appenix available at: N U M B E R 32, S P R I N G 2017

13 Chart 5: Use of Quality Ajustment for Government Input an Output Measurement in OECD Countries e e e e / K K K e E Y < E / K e W ^ W D Note: The countries for not measure inclue: Austria, Belgium, Canaa, Costa Rica, Czech Republic, Estonia, Italy, Japan, Latvia, Luxembourg, Mexico, Norway, Slovenia, Spain, Sween an Switzerlan. Don t know is comprise of countries: South Africa, Netherlans, Israel an Finlan. No quality ajustment for input or output ata countries inclue: Portugal, Polan, Germany, France, Denmark, Chile an Austria. While input ata only is New Zealan an output ata only countries inclue: UK, Slovak Republic, Irelan an Hungary. Source: 2016 OECD Survey on Measures of Prouctivity in OECD an Partner Countries. taught; other factors, such as the income level an eucational attainment of the parents, the quality of teachers, an class size, also having an impact. This is the calle the attribution problem that is associate with ientifying the contribution of outputs to outcomes in the public sector. The OECD survey on public sector prouctivity measurement aske countries whether they are carrying out explicit quality ajustment (i.e. applying a numerical correction to the level of outputs base on the resulting outcome(s)) to their input an output measures. Quality ajustment oes not constitute a wiesprea practice among OECD countries (Chart 5). Only 5 of the 16 countries that measure public sector prouctivity report making use of quality ajustment. Hungary, Irelan, the Slovak Republic an the Unite Kingom apply quality ajustments to output ata, while New Zealan reports using quality ajustment for input measures. Future Steps in Government Prouctivity Measurement The examples of countries that have unertaken steps to measure prouctivity in the whole public sector show that while the task is very ifficult an complicate, it can be one. However, without proper output an input volume measurements, prouctivity cannot be calculate. Future OECD work coul focus on the following five areas: (i) improvements to input meas u r e m e n t a n c o s t a c c o u n t i n g ; ( i i ) stanarization an comparability of measures; (iii) output measurement beyon the eucation an health sectors; (iv) a typology of activities at the micro level; an (v) intra-governmental coorination on prouctivity measurement. The stanars for input measurement are better evelope than for output measurement, although even in this area more work can be one. The key to measuring prouctivity in the public sector at any of the three levels macro, meso an micro is the existence of a reliable I N T E R N A T I O N A L P R O D U C T I V I T Y MON I T O R 192

14 cost accounting system that is able to separate out the input costs to the various outputs. Different steps can be taken to move this agena forwar: i) harmonizing accounting methos in the public sector, in orer to ensure consistency in measuring costs; ii) esigning new public sector financial IT systems not only as accounting tools, but as enablers for prouctivity measurement; iii) using feeback/knowlege of public managers to unerstan which cost ata are meaningful an useful to improve prouctivity in their specific area of work. Much more effort can be unertaken with regars to output measurement. Given that it is easier to ientify the final outputs for iniviual services for iniviuals, it is most evelope for eucation, health care an social care services. Next steps to be taken inclue broaening output measurement to other iniviual services. Developing methoologies to measure collective service outputs still seems a rather istant goal an will take a consierable amount of resources an time. Mixe methos using the output metho for some key services an the input metho for the remainer can provie interim solutions. A stanarization of measurement practices woul also be useful to enhance comparability. International organizations, such as the European Commission an the OECD, have contribute to eveloping a methoology for price an volume measurement in national accounts. The work is the most avance for the health care an eucation services. In orer to compare prouctivity in these sectors across countries, further stanarization of input an output measurement is neee. This might require some countries to change the inicators they currently use, e.g. from the number of pupils to the number of pupil hours taught. There is little known about how countries measure prouctivity at the microlevel, i.e. the organization/function level. Most frequently, existing micro level measurements exist in the eucation an health area in the form of league tables comparing the performance of iniviual schools an hospitals. However, the measurement of prouctivity or value for money of various government functions woul be equally beneficial. There is great interest from countries in this microlevel measurement an great scope for learning from each other. The UK example (see Box 2) shows how this coul be carrie out. The OECD Secretariat is planning case stuies to look at how prouctivity/value for money can be measure for the procurement function an for igitalization. Finally, intra-governmental co-orination is essential to avance measurement efforts. Institutional responsibilities for measuring public sector prouctivity are ifferent across countries. Frequently, efforts are le by National Statistical Offices with the involvement of line ministries, who are in the best position to ecie on the most important final outputs to be counte in their area. The assignment of clear responsibility for public sector prouctivity measurement to a specific institution an proviing it with aequate resources to fulfil this task can help clarify roles an facilitate co-orination of ata collection. Prouctivity Commissions, as for example in Australia an New Zealan, or Prouctivity Boars have measure prouctivity in specific public sector omains (Dougherty an Rena, 2017), an coul also play a useful role in co-orinating an promoting prouctivity measurement in the public sector. Conclusion The rivers of public sector performance an prouctivity are manifol. While this presents a number of challenges, it also means that governments can mobilize various tools an improve ifferent processes to increase public sector prouctivity. This article lays the grounwork for 193 N U M B E R 32, S P R I N G 2017

15 further OECD research to better unerstan public sector prouctivity an how it can be improve. It suggests several ways forwar to aress measurement challenges an to elve eeper into how governance frameworks an processes can be mobilize to achieve greater public sector prouctivity. As a first step, efforts shoul be mae to accurately measure the prouctivity of the public sector to help focus attention on public sector prouctivity as a policy objective, to support benchmarking an to ientify an compare possible strategies to improve prouctivity. Further evelopment an stanarization of measures of government output an the inputs use to generate them beyon the eucation an health sectors are worthwhile. Furthermore, there is great value an interest in assessing public sector prouctivity in greater etail on the microlevel, i.e. evaluating closely the prouctivity of iniviual government organizations an functions. Aitional research on the policy rivers for public sector prouctivity is also require. These inclue implementing igital government strategies, strategic human resource management, creating an enabling environment conucive to innovation, an unerstaning better the impact of bugeting an regulatory practices. New approaches shoul be able to consier the complexity of trae-offs an tensions. An example is associate with the increasing uptake by public sectors of new technologies that go beyon pure financial ecisions, such as convenience an personal ata protection, or leveraging the economies of scale inherent to clou computing. An there are ways to assess the impact on prouctivity an public value creation brought about by new forms of eploying technology (e.g. share ICT services an how they contribute to public sector prouctivity) or sharing an processing ata. Only by better unerstaning the internal workings of the 'black box' by which governments transform inputs into outputs an ultimately public outcomes can we map the rivers of prouctivity, promote the iffusion of innovations to strengthen public sector prouctivity an improve public sector effectiveness. This task is mae more ifficult by the rising complexity of government relationships an partnerships, in a context of co-esign an coelivery. Bounary crossing activities an effects (for example, when prouctivity increases take place through the collaboration of organizations, sectors an citizens) are frequently not accounte for in current measurement systems. So while innovation is certainly a vector for improving prouctivity in the public sector, as it is in the private sector, it also represents a measurement challenge as innovation projects in general can have multiple actors, at both the same an ifferent levels of the public sector. Thus it can be ifficult to isolate the specific impact an effect of particular project an/or particular organisation within a project. References Castelli A, Dawson D, Gravelle H an Street A (2007) Improving the Measurement of Health System Output Growth, Health Economics, Vol. 16, No. 10, pp Department for Communities an Local Government (DCLG) (2015) Public Sector Efficiency: Final Report from the cross-whitehall analysts' group July. Strategic Analysis Team, Analysis an Innovation Directorate. mimeo, PPT presentation. Diewert, W.E. (forthcoming) "Prouctivity Measurement in the Public Sector: Theory an Practice," Oxfor Hanbook on Prouctivity. Dunleavy, P. an L. Carrera (2013) Growing the Prouctivity of Government Services, Ewar Elgar Publishing. Dunleavy, P. (2016) "Public Sector Prouctivity - Measurement Challenges, Performance Information an Prospects for Improvement," November _Public_Sector_Prouctivity_- _Measurement_Challenges_Performance_Infor mation_an_prospects_for_improvement I N T E R N A T I O N A L P R O D U C T I V I T Y MON I T O R 194

16 Dougherty, S. an A. Rena (2017) "Pro-Prouctivity Institutions: Learning from National Experience," International Prouctivity Monitor, Spring, Vol. 32. European Communities, International Monetary Fun, Organisation for Economic Co-operation an Development, Unite Nations an Worl Bank (2009) System of National Accounts 2008, Available at: nationalaccount/ocs/sna2008.pf. European Union (2016) Hanbook on Prices an Volume Measures in National Accounts, Eurostat, / /KS-GQ EN-N.pf/ b e47b798b2. Gutacker N., C. Bojke, S. Daione, N. Devlin, D. Parkin an A. Street (2011) Truly Inefficient or Proviing Better Quality of Care? Analysing the Relationship Between Risk-Ajuste Hospital Costs an Patients' Health Outcomes Centre for Health Economics. University of York. Lau, Ewin (2017) Unerstaning Public Sector Prouctivity, OECD Working Paper Series, forthcoming. Lorenzoni, Luca (2015) Note on Price an Volume Measures for Health Care, Document prepare for the OECD Working Party on National Accounts, CSSP/WPNA(2015)14&ocLanguage=En. Moretti, D. (2016) "Accrual Practices an Reform Experiences in OECD Countries Results of the 2016 OECD Accruals Survey," OECD Journal on Bugeting, Vol. 16, No. 1. Netherlans Institute for Social Research (2015) The Netherlans Public Sector Achievement in 36 Countries, Publications_2015/ Public_sector_achievement_in_36_countries. National Health Services Information Centre (2011) Provisional Monthly Patient Reporte Outcome Measures (PROMs) in Englan. DOI: x.oi.org/ / en OECD (2010) Improving Health Sector Efficiency: The Role of Information an Communication Technologies. OECD Health Policy Stuies, OECD Publishing, Paris en OECD (2013) Government at a Glance 2013, OECD Publishing, Paris. OECD (2015a) Builing on Basics, OECD Publishing, Paris. OECD (2015b) Tax Aministration 2015: Comparative Information on OECD an Other Avance an Emerging Economies, OECD Publishing, Paris. OECD (2015c) The State of Public Finances OECD (2016) OECD Compenium of Prouctivity Inicators 2016, OECD Publishing, Paris. Office for National Statistics (2005) Atkinson Review: Final report. Measurement of Government Output an Prouctivity for the National Accounts. Robano, Virginia (2016) "Measuring Public Sector Prouctivity," Backgroun ocument prepare for the OECD. Schreyer P. an M. Mas (2013) Measuring Health Services in the National Accounts: an International Perspective, Schreyer, P. (2010) Towars Measuring the Volume Output of Eucation an Health Services: A Hanbook," OECD Statistics Working Papers, 2010/02, OECD Publishing. DOI: / 5km34g1zk9x-en Statistics Denmark (2013) General Government Output an Prouctivity, Site/Dst/Ugivelser/GetPub- File.aspx?i=18683&si=gengov. 195 N U M B E R 32, S P R I N G 2017