Experimental hospital multifactor productivity estimates

Productivity measures are useful to assess the performance and efficiency of resource use. The ABS currently compiles multifactor productivity estimates for market sector industries but not for non-market sector industries such as health care.¹ Non-market sector industries have a large portion of output provided at prices that are not economically significant; that is, where goods and services are provided to final consumers at prices below the cost of provision, such as public hospital services.

Given the importance of non-market industries to the Australian economy, the ABS has a research agenda to address this gap in productivity statistics. This paper presents experimental estimates of multifactor productivity for hospitals, a sub-division of the health and social assistance industry.

The estimates show that between 2008-09 and 2018-19, hospital labour productivity grew on average 0.5% per annum. Increased use of intermediate inputs per hour worked contributed about 0.3 percentage points per annum. The average annual growth rate of multifactor productivity was 0.1%. Labour input was a main driver for output growth, contributing about 1.8 percentage point to annual growth of 3.5% in output. This reflects that hospital output is labour intensive. 

It is important to note that these estimates do not reflect the impact of COVID-19, as the analytical timespan ends at 2018-19. However, the methodology used in this paper will reveal medium to long term impacts of COVID-19 on hospital output and productivity when data becomes available.

The ABS welcomes comments and suggestions from readers. To provide feedback, please email economic.research@abs.gov.au.

Hospital output plays an important role in the Australian economy. According to the Australian Government’s 2021 Intergenerational Report, the impact of an ageing population, technological advancement and rising incomes is expected to increase Commonwealth health expenditure from 19% of total government spending in 2021-22 to 26% in 2060-61. Commonwealth funding for public hospitals is expected to double between 2020-21 and 2031-32.² 

Hospitals are classified under the Australian and New Zealand Standard Industrial Classification (ANZSIC) Subdivision 84 in Industry Division Q (Health care and social assistance). Hospitals are significant to Division Q as hospital services comprise over one third of the industry’s output. Australian hospitals employed around 597,000 people in 2018-19.³

The Australian hospital system consists of public and private hospitals. Public hospitals deliver around 60% of hospital separations in admitted patient care, as well as most emergency department activity and non-admitted patient care.⁴

The ABS classifies public hospitals as non-market producers because the majority of their output is provided to patients free of charge, or at prices which are not economically significant.  In contrast, private hospitals are considered to be market producers, though some private hospitals are contracted to provide public hospital services and provide services directly to public patients.

Multifactor productivity is calculated as the ratio of output growth (in volume terms) to growth of a combined input (volume) measure of labour, capital, and intermediate inputs (i.e. goods and services consumed in production).⁵ The output measure is based on the ABS’s recently published output index for hospitals.⁶ Input measures were based on the Labour Account hours worked data,⁷ newly constructed experimental capital services index for the health industry,⁸ and intermediate input indexes constructed from administrative data and the Government Finance Statistics (GFS) dataset.

The ABS recently published experimental volume indexes for hospital output, which enhances coverage compared to the existing volume index used in the National Accounts.⁶ The experimental index includes public hospital emergency department activity and non-admitted care. This coverage enhancement also enables measurement of shifts across categories of care for public hospitals over time. For private hospitals, the index covers admitted patient care only.

The inclusion of non-admitted patient care in the index is especially important with an increasing shift from admitted care to non-admitted care as the result of advances in medical technology (Gu & Morin 2014). Such shifts can be regarded as efficiency gains as less labour is used in treating patients in an outpatient clinic than in a hospital ward.

The approach underpinning these experimental indexes is the direct volume index approach, where an annually chained Laspeyres index was calculated by combining output quantity indicators and cost weights.

Although private hospitals are regarded as market producers, the same direct volume index approach was used.  A deflation approach would require detailed price indexes matching the level of disaggregation of output quantity, to control for quality change. In practice, such price indexes were not available. Therefore, the direct volume index approach was preferred. It is noted that US Bureau of Labour Statistics also used the direct volume index approach to measuring output of private hospitals (Chansky, et al 2015).⁹

Table 1 shows the quantity indicators used.

Quantity and cost data were sourced mainly from the Australian Institute and Welfare (AIHW) and Independent Hospital Pricing Authority (IHPA).

A volume index for the hospital sub-industry as a whole was estimated as aggregation of the private and public hospital output indexes. The weight for public hospital output was calculated from public hospital expenditure via the GFS dataset.¹⁰ The weight for private hospital output was based on private hospital revenue, sourced from the ABS Australian Industry dataset.¹¹ ¹²

Figure 1 shows volume indexes for public hospital output, private hospital output, and total aggregate hospital output over the period from 2004-05 to 2018-19. Aggregate hospital output growth tracks closely to public hospital output growth because the public hospital sector is larger.

The ABS currently compiles productivity estimates using ABS Labour Account hours worked as the labour input measure. For hospitals, ANZSIC Subdivision 84 total annual hours worked were used.   The Labour Account estimates are aligned with the production concept in the National Accounts in terms of both coverage and methodology.¹³

The ABS also compiles quality adjusted hours worked in the official productivity statistics to take into account change in labour quality over time.¹⁴ However, quality adjusted hours worked are only available at the ANZSIC division level. In this paper, a simplified method of stratification by occupation was used. For example, medical doctors are considered as having different level of education and skill sets compared to nurses.¹⁵ The relative proportion between doctors and nurses can change over time.

 A quality adjusted index of hours worked for hospitals was derived by splitting total number of hours worked between public hospitals and private hospitals according to their relative proportion in full time equivalent (FTE) employment. This derivation assumed there was no difference in average hours worked per FTE between public and private hospitals. Public hospital FTE were sourced from the Australian Institute of Health and Welfare (AIHW) publication of hospital resources.¹⁶ Private hospital FTE were sourced from the ABS publication – Private Hospitals.¹⁷

Employed staff was disaggregated into five broad categories of occupations:

• salaried medical officers;
• nurses;
• diagnostic and allied health professionals;
• administrative staff and personal care staff; and
• other.

The proportion of hours worked for each category was approximated by the relative proportion of FTE employment. A labour composition index was calculated as a Törnqvist index based on FTE weighted by wages. Wages for public hospitals were sourced from the AIHW, however wages data for private hospitals were not available, so wages for public hospitals were also used for relative weights for private hospitals.¹⁸ The composition indexes for public and private hospitals were aggregated to total composition index based on their respective total labour costs.

A quality adjusted index of hours worked was calculated as the product of an index of hours worked and the labour composition index (see Appendix A for details of the method).

Figure 2 compares three different labour input indexes: an unadjusted index of total hours worked, quality adjusted hours worked, and an index based on the sum of FTE employment. The three indexes followed similar trends, though the FTE index generally grew at a faster rate. The hours worked index increased by an average of 3.0% annually over the period from 2008-09 to 2018-19, compared to 3.4% for the FTE index. This means that average hours worked per FTE has decreased.

The quality adjusted hours worked index increased at an average annual rate of 3.2% compared to 3.0% for the unadjusted index of total hours worked. Higher growth in the quality adjusted hours worked was due to an increasing proportion of salaried medical staff with higher wages in the staff FTE composition.

The ABS only currently measures capital services (for purposes of compiling multifactor productivity statistics) for market sector industries. Divisions O (Public administration and safety), P (Education and training) and Q (Health care and social assistance) are excluded.

Experimental capital services indexes have been constructed for the health care industry as a whole (Division Q). The methodology for the new capital services indexes is described in a separate paper,⁸ and these indexes are used in this paper as a proxy for growth in hospital capital services.

A limitation of using the Division Q capital services index for hospitals is that the composition of assets providing capital services in hospitals may differ from the average of the health care industry. However, the impacts on productivity measures are expected to be immaterial, given hospital activity is labour intensive.

Intermediate inputs are goods and services consumed in the production process such as energy, materials and services. While estimates for intermediate inputs for ANZSIC Division Q are available from the National Accounts, they also include intermediate inputs consumed by non-hospital units of the health care industry, which are outside the scope of this paper. For this analysis, estimates for only the hospital component of the industry were constructed.

For public hospitals, total current price expenditure on intermediate inputs were estimated using GFS expenditure data for hospitals. This expenditure was then disaggregated into categories according to AIHW public hospital expenditure data (e.g. AIHW 2019).¹⁹ For each expenditure category, volume growth was estimated by deflating current price expenditure by an appropriate price index. An annually chained Laspeyres volume index was then calculated by aggregating across all the expenditure categories.

For private hospitals, total current price expenditure on intermediate inputs were sourced from the ABS Australian Industry publication.¹² These were then disaggregated into expenditure categories using the ABS Private Hospital Statistics publication.²⁰ An annually chained Laspeyres volume index was then calculated by aggregating across all the expenditure categories in the same way as for public hospitals.

A total volume index for intermediate input for all hospitals was calculated as a Laspeyres index by aggregating the public hospital and private hospital indexes.²¹

The methodology for calculating the intermediate input indexes is described in Appendix B.

An aggregate input of labour inputs, capital inputs and intermediate inputs was constructed as a Törnqvist index using cost shares as weights. The cost share for labour is compensation of employees (COE) divided by the current price value of gross output. The cost share for capital is gross operating surplus (GOS) divided by the current price value of gross output.²²  The cost share for intermediate inputs is the ratio of expenditures on intermediate inputs to current price value of gross output.

COE and GOS are only compiled at ANZSIC division level in the National Accounts, therefore cost shares for public hospitals were estimated using GFS data. As public hospitals are classified as non-market producers, GOS is set equal to consumption of fixed capital (COFC).²³

In theory, the weight for capital should include opportunity costs (return of investment on capital) regardless of whether a producer is market or non-market. Setting GOS to equal COFC may underestimate the contribution of capital services to public hospital output.²⁴

Cost shares for private hospitals were estimated in the same way as for market producers where GOS is estimated based on sales revenue and operating expenses. The revenue and expenses data were sourced from the ABS Australian Industry.

Figure 3 shows the average cost shares for each component of public hospitals, private hospitals, hospital as a sub-industry, and the total health care industry (Division Q).²⁵ The capital cost weight for public hospitals is considerably lower than that for private hospitals and Division Q as a whole. In comparison, the capital cost weight for private hospitals is similar to Division Q. It is worth noting that private hospitals have a lower labour cost share and higher intermediate input cost share.

The total input volume index and the hours worked index grew at similar rates (Figure 4), reflecting that hospitals are labour intensive, with labour comprising around 60% of total inputs (Figure 3). The total aggregate input index based on hours worked rose by an average of 3.4% per year, compared to 3.0% for the hours worked index and 3.8% for the intermediate input index. The total aggregate input index based on quality adjusted hours worked increased by an average of 3.5% per year, due to shifts to higher quality labour over time.

Figures 5 and 6 summarise the productivity estimates.  

Between 2008-09 to 2018-19, labour productivity for the hospital sub-industry grew by an average of 0.5% per year. The average annual growth rate of labour productivity between 2008-09 and 2013-14 was 0.8%.

Labour productivity based on the quality adjusted hours worked grew at an average annual rate of 0.3%. Compositional change in labour input accounted for about 0.2 percentage points of annual growth in labour productivity per year on average, compared to labour productivity growth estimated on the unadjusted hours worked basis.

Hospital multifactor productivity increased by an average of 0.1% per year, with most of the growth occurring in earlier periods. The average annual growth rate of multifactor productivity between 2008-09 and 2013-14 was about 0.7%.

Capital productivity fell by 1.3% annually on average. The declining trend in capital productivity was driven by rapid growth in capital services for the health care industry. Because the overall weight for capital input was small, this decline only had minor impacts on multifactor productivity.  Capital services grew faster than hours worked, therefore hospitals generally experienced capital deepening, i.e. an increase in use of capital services per unit hour worked.

Figure 7 shows a breakdown of sources for growth in labour productivity (calculated in terms of hours worked). In terms of log growth, increased use of intermediate inputs per hour worked contributed about 0.3 percentage points of growth on average annually, while capital deepening contributed an annual average 0.1 percentage point.

Figure 8 shows that labour input was the main driver for growth in hospital output, contributing 1.8 percentage points annually on average, compared to a 1.3 percentage point contribution from intermediate inputs.

Figure 9 shows that growth in labour input accounted for more than half of total cumulative output growth over the period from 2008-09 to 2018-19.

There are some limitations which could affect the interpretation of the results, in addition to limitations discussed in an earlier paper on hospital output measures.⁶

The hospital output measure used in this analysis have not been explicitly quality adjusted from the patient’s perspective. Improvements in treatments over time that deliver quality-related benefits to patients may not be fully captured. Although some National Statistical Offices have experimented in this area, there is no internationally agreed framework for incorporating explicit quality adjustments in output measures from a national accounts perspective.²⁸

To estimate hospital intermediate input indexes in volume terms, experimental deflators were developed. Ideally, input producer price indexes should be used to deflate each component of these deflators, but these were not available in all cases. See Tables B1 and B2 in Appendix B for details of price indexes used.

Capital services were measured using the Division Q capital services index. A limitation of using the Division Q capital services index for hospitals is that composition of assets providing capital services in hospitals could differ from the average of the health care industry. However, as mentioned earlier, the impacts of such differences on productivity measures are expected to be immaterial, due to the small capital cost share. Nonetheless, for completeness, it would be desirable to measure capital services using data specific to hospitals (ABS 1997).

Hospital productivity growth has generally been lower than Australian productivity growth, although health care is an area of the economy where there has been technological progress (such as the development of new or improved procedures and new drugs in treating patients). Labour productivity grew at an annual rate of 0.5% over the period from 2008-09 to 2018-19, lower than the market sector average. Increased use of intermediate inputs (per hours worked) was a major contributor to labour productivity growth.

Multifactor productivity grew at an even slower rate of about 0.1% per year. This means the contribution of multifactor productivity to hospital output was relatively minor, with the majority of output growth attributed to growth in labour input. This reflects the fact that hospital services are labour intensive.

Improving output measures for hospitals will continue to be a challenging area. Future work could include refinement of output indexes by using a finer level of stratification and allowing for the presence of comorbidities. Other possibilities include constructing indexes from an integrated dataset covering all three categories of patient care, experimenting with explicit quality adjustment, and extending the time series to incorporate the full time span of labour accounts data.

The multifactor productivity estimates could be improved by developing a capital services index specifically for hospitals.

This paper was authored by Qinghuan Luo and Jason Annabel. The authors would like to thank Derek Burnell, Katherine Keenan and Jacqui Jones for their comments and input.

Australian Institute of Health and Welfare (AIHW) (2019) Hospital resources 2017-18: Australian hospital statistics. 

Australian Bureau of Statistics (ABS) (1997), Measuring outputs, inputs and productivity for Australian public acute care hospitals.

Australian Bureau of Statistics (ABS) (2015) Australian System of National Accounts Concepts, Sources and Methods.

Chansky B, Garner C A, Raichoudhary R (2015), New measure of labour productivity for private community hospitals: 1993–2012, Monthly Labour Review, U.S. Bureau of Labour Statistics, October 2015.

Gu W, Morin S (2014), Experimental measures of output and productivity in the Canadian hospital sector, 2002 to 2010, in Measuring Economic Sustainability and Progress, edited by D.W. Jorgenson, J.S. Landefeld and P. Schreyer, 575-594. Chicago: University of Chicago Press.

Office of National Statistics (ONS) (2012), Quality adjustment of public service health output: current method, ONS.

This appendix describes the method for estimating volume indexes of hospital intermediate inputs.

Total current price expenditure on intermediate inputs for public hospitals was estimated from the GFS. The total expenditure was disaggregated using the administrative data on hospital expenditure from the AIHW Publication of Hospital Resources. Table B1 summarises expenditure categories and their average proportions from 2014-15 to 2018-19. The third column lists price indexes/IPDs used as deflators.

Total current price expenditure on intermediate inputs for private hospitals was sourced from the ABS Australian Industry. The total expenditure was disaggregated using private hospital expenditure categories derived from the ABS Private Hospitals (Table B2). As the private hospital expenditure data does not have a split between drug supplies, and medical and surgical supplies, the split between these two categories was based on public hospitals.

Similarly, an intermediate input volume index for private hospitals can be calculated using equation (B1).