Adjusting productivity statistics for
variable capacity utilisation
Working harder or hardly working?
Statistics New Zealand Working Paper No 12–02
Adam Tipper and Nicholas Warmke
Crown copyright ©
This work is licensed under the Creative Commons Attribution 3.0 New Zealand licence. You
are free to copy, distribute, and adapt the work, as long as you attribute the work to Statistics
NZ and abide by the other licence terms. Please note you may not use any departmental or
governmental emblem, logo, or coat of arms in any way that infringes any provision of the
Flags, Emblems, and Names Protection Act 1981. Use the wording 'Statistics New Zealand' in
your attribution, not the Statistics NZ logo.
Disclaimer
The Statistics New Zealand Working Paper series is a collection of occasional papers on a
variety of statistical topics written by researchers working for Statistics New Zealand. Papers
produced in this series represent the views of the authors, and do not imply commitment by
Statistics NZ to adopt any findings, methodologies, or recommendations. Any data analysis
was carried out under the security and confidentiality provisions of the Statistics Act 1975.
Liability statement
While all care and diligence has been used in processing, analysing, and extracting data and
information in this publication, Statistics New Zealand gives no warranty it is error free and will
not be liable for any loss or damage suffered by the use directly, or indirectly, of the
information in this publication.
Citation
Tipper, A, & Warmke, N, (2012). Adjusting productivity statistics for variable capacity
utilisation: Working harder or hardly working? (Statistics, New Zealand Working Paper No
12–02). Available from www.stats.govt.nz.
ISBN 978-0-478-37791-0 (online)
ISSN 1179-934X (online)
Published in October 2012 by
Statistics New Zealand
Tatauranga Aotearoa
Wellington, New Zealand
Contact
Statistics New Zealand Information Centre
info@stats.govt.nz
Phone toll-free 0508 525 525
Phone international +64 4 931 4610
www.stats.govt.nz
Contents
List of tables and figures ................................................................................................... 4
Abstract ............................................................................................................................... 6
Acknowledgements.......................................................................................................... 6
1 Introduction .................................................................................................................... 7
2 Background .................................................................................................................... 9
Theory .............................................................................................................................. 9
Measuring capital........................................................................................................... 10
Productivity measurement and interpretation................................................................ 11
3 Availability of data for adjusting for variable capacity utilisation ......................... 13
New Zealand Institute of Economic Research’s capacity utilisation index ................... 13
New Zealand Institute of Economic Research’s capacity as a constraint series .......... 16
Relationship between CUBO and CAAC....................................................................... 17
4 Applying capacity utilisation measures to capital assets ...................................... 19
5 Empirical analysis ....................................................................................................... 20
Impact on measured sector input .................................................................................. 20
Impact on measured sector MFP .................................................................................. 22
6 Summary....................................................................................................................... 23
References ........................................................................................................................ 24
Appendix: Industry effects .............................................................................................. 25
Manufacturing ................................................................................................................ 25
Construction ................................................................................................................... 26
Wholesale trade ............................................................................................................. 27
Retail trade..................................................................................................................... 28
Accommodation, cafes, and restaurants ....................................................................... 29
Transport and storage ................................................................................................... 30
Finance and insurance .................................................................................................. 31
Business services .......................................................................................................... 32
3
List of tables and figures
Tables by chapter
3 Availability of data for adjusting for variable capacity utilisation
1 Capacity utilisation across growth cycles ................................................................... 14
2 Correlation matrix for industry-level capacity utilisation measures ............................ 18
5 Empirical analysis
3 Summary of capacity adjusted measures and industries........................................... 20
Figures by chapter
2 Background
1 Capacity output and costs ............................................................................................ 9
2 Theoretical utilisation of capital, capital stock available and capital stock utilised .... 11
3 Actual utilisation of capital, capital stock available and capital stock utilised ............ 11
3 Availability of data for adjusting for variable capacity utilisation
4 Capacity utilisation and GDP, annual percentage change......................................... 14
5 Capacity utilisation rates, combined and industry specific CUBO series .................. 15
6 Manufacturing, capacity utilisation, capital input, and output growth ......................... 15
7 Output growth and capacity utilisation growth, construction ...................................... 16
5 Empirical analysis
8 Measured sector capital services, with and without capacity adjustments, annual
percentage change ................................................................................................... 21
9 Measured sector capital productivity, with and without capacity adjustments,
annual percentage change ....................................................................................... 21
10 Measured sector total inputs, with and without capacity adjustments, annual
percentage change ................................................................................................... 22
11 Measured sector MFP, with and without capacity adjustments, annual
percentage change ................................................................................................... 22
Appendix: Industry effects
Manufacturing: MFP with and without capacity adjustment, indexes ........................... 25
Manufacturing: MFP with and without capacity adjustment, annual percentage
change ...................................................................................................................... 25
Construction: MFP with and without capacity adjustment, indexes .............................. 26
Construction: MFP with and without capacity adjustment, annual percentage
change ...................................................................................................................... 26
Wholesale trade: MFP with and without capacity adjustment, indexes ........................ 27
Wholesale trade: MFP with and without capacity adjustment, annual percentage
change ...................................................................................................................... 27
4
Adjusting productivity statistics for variable capacity utilisation: Working harder or hardly working?
Retail trade: MFP with and without capacity adjustment, indexes ................................ 28
Retail trade: MFP with and without capacity adjustment, annual percentage
change ...................................................................................................................... 28
Accommodation, cafes, and restaurants: MFP with and without capacity
adjustment, indexes .................................................................................................. 29
Accommodation, cafes, and restaurants: MFP with and without capacity
adjustment, annual percentage change ................................................................... 29
Transport and storage: MFP with and without capacity adjustment, indexes............... 30
Transport and storage: MFP with and without capacity adjustment, annual
percentage change ................................................................................................... 30
Finance and insurance: MFP with and without capacity adjustment, indexes.............. 31
Finance and insurance: MFP with and without capacity adjustment, annual
percentage change ................................................................................................... 31
Business services: MFP with and without capacity adjustment, indexes ..................... 32
Business services: MFP with and without capacity adjustment, annual percentage
change ...................................................................................................................... 32
5
Abstract
This paper outlines the implications of adjusting productivity statistics for a variable rate of
capacity utilisation of capital. Capacity utilisation data from the New Zealand Institute of
Economic Research shows that capacity utilisation is not constant over time. Capacity
utilisation adjustment leads to marginally lower capital input growth and higher multifactor
productivity (MFP) growth at the measured sector level.
The paper concludes that capacity utilisation adjustment has minimal impact on long-term
growth, leading to marginally lower capital input growth and higher MFP growth at the
measured sector level. In the short term, the effects of adjusting productivity statistics for
variable capacity utilisation are more significant, leading to less volatile MFP estimates.
Keywords
Productivity measurement; industry; capital; capacity utilisation; multifactor productivity
Acknowledgements
The authors thank Peter O’Connor (New Zealand Institute of Economic Research) for
supplying the capacity utilisation data, Toby Hunter (formerly of Statistics New Zealand)
for help on an earlier version of the paper, and participants at the 52nd New Zealand
Association of Economists Conference, Wellington, in July 2011.
6
1 Introduction
Productivity is a measure of how efficiently inputs (such as machinery, computer
software, land, and labour) are being used within the economy to produce outputs.
Productivity is commonly defined as a ratio of a volume measure of output to a volume
measure of input, that is:
Productivity = Output / Input
Growth in productivity means that over time, a nation or an industry can produce more
output from the same amount of inputs, or the same amount of output with fewer inputs.
Observed productivity growth can reflect changes in efficiency (getting more from given
inputs), technological change, or measurement error. Productivity measures are vital to a
better understanding of long-term improvements in New Zealand's living standards,
economic performance, and international competitiveness.
The factors of production, namely, capital and labour, need to be accurately measured to
meaningfully reflect their contribution to the production process. Labour is relatively easy
to understand and measure through either hours of work or other measurements of
employment (eg full-time equivalent employees). Measuring the contribution of capital to
the production process is much more difficult. There is no easy way to measure ‘machine
hours’ for capital used in the production process. Because of this it is assumed that the
capital services available for use in production are proportional to the volume of
productive capital stock. Productive capital stock represents the total capital stock,
adjusted for a decline in efficiency.
In the absence of appropriate data on utilisation rates, Statistics NZ (and other
international statistical agencies) assumes capital and labour are used at a constant rate
over time when estimating capital and multifactor productivity (MFP). This is
recommended by the Organisation for Economic Co-operation and Development’s
(OECD) 2001 manual Measuring Productivity: Measurement of Aggregate and Industry
Level Productivity Growth. Doing this facilitates international comparability of productivity
statistics. However, in reality, it is acknowledged that utilisation of inputs may not be
constant.
Variable utilisation of labour will largely be reflected in existing data (Hulten, 2001) due to
the way labour data is collected and compiled. Changes in the intensity of use of labour
inputs will also to some extent be captured in the ‘hours paid’ measure of labour input (ie
through paid overtime). However, with no available measure of capital input utilised in
production, capital services are assumed to be utilised at a constant rate.
Variable capacity utilisation may lead to a gap between actual and measured productivity
and may be one possible reason for the procyclicality of MFP. If input cyclicality is
routinely unaccounted for, measured productivity may be procyclical even if actual
productivity does not change. Therefore, addressing the assumption of a constant rate of
capacity utilisation is vital to verifying movements in capital and MFP.
This paper assesses the impact and implications of applying a variable rate of capacity
utilisation to capital stock data used in productivity measures.
7
Adjusting productivity statistics for variable capacity utilisation: Working harder or hardly working?
Outline of this paper
This paper begins by outlining the motivations for adjusting for a variable rate of capacity
utilisation, the methodology and assumptions for productivity measurement, and the
implications of these assumptions for productivity growth. It proceeds with review of the
potential data sources that can be used to adjust capital inputs for variable utilisation
rates. The empirical impact on New Zealand’s measured sector capital and MFP growth
is then discussed.
At the measured-sector level the impact of adjustment is minimal but evident, especially
during turbulent years in the economy. The impact on individual industries that comprise
the measured sector varies. The paper concludes that adjusting productivity statistics for
variable capacity utilisation leads to smoother productivity estimates in the short-term but
has minimal impact on long-term growth.
8
2 Background
Theory
Capacity utilisation reflects the difference between the potential and actual use of an
input. Utilisation is highest when full use is made of labour and capital. In a static context,
capacity utilisation can refer to either ‘engineering capacity’ or ‘economic capacity’
(Shaikh & Moudud, 2004). Engineering capacity is the maximum sustained production
that is possible over a period; that is, the physical potential of using inputs. On the other
hand, economic capacity refers to the desired level of output from inputs. This definition
takes account of the cost of additional time units of capital or labour. In this framework,
Berndt and Morrison (1981) define capacity output as the minimum point on the short-run
average cost curve, where it is tangential to the long-run average cost curve; increasing
output beyond capacity output is possible but brings cost pressures (see figure 1).
In the long-term, full capacity may be equated with the firm’s optimal long-run equilibrium
point, though capacity utilisation may vary over time. Low levels of capacity utilisation
indicate slowing economic activity while high levels which suggest strong activity and
therefore inflationary pressures (Hornstein, 2002). Nelson (1989) shows that, while the
level of utilisation may differ slightly, the economic and engineering capacity measures
are related empirically, but that relationship may change over time.
Figure 1
1 Capacity output and costs
Capacity output and costs
Basu (1996) notes the gap between actual and measured productivity most likely comes
from cyclical errors in measuring inputs, such as unobserved changes in capital utilisation
or the intensity of work effort. Variable capacity utilisation is one possible reason for the
procyclicality of MFP (Basu & Fernald, 2000) as measures of capacity utilisation tend to
be procyclical. If input cyclicality is routinely unaccounted for, measured productivity may
9
Adjusting productivity statistics for variable capacity utilisation: Working harder or hardly working?
be procyclical even if actual productivity does not change. Therefore, addressing the
assumption of a constant rate of capacity utilisation is vital to verifying movements in
capital and MFP.
However, mismeasurement of capital services is not the only possible explanation for
procyclical MFP. Technology shocks (which are reflected in MFP) may lead to output and
labour input growth and therefore drive growth cycles. This is the dominant assumption
underlying real business-cycle theory. Alternatively, procyclical MFP may result from
increasing returns to scale; that is, the economy becomes more efficient by moving to
higher levels of activity. As such, procyclical MFP may still be observed even when
variable capacity utilisation is accounted for.
Measuring capital
As mentioned earlier, measuring capital as an input into the production process is not an
easy task. With labour, businesses and individuals can be surveyed to account for the
hours of work that feed into the production process. Administrative data sources, such as
tax data or business data can be accessed to ascertain how many people work in an
industry or in the economy. In contrast, capital has no readily available data source on
the number of hours capital is in operation, or how much it contributes to the production
process in an easy-to-understand form. In considering the contribution of assets to a
production process, it is the volume of the capital services produced by these assets that
is relevant and not the stock of the assets.
Statistics NZ uses a perpetual inventory method (PIM) to derive the productive stock of
capital by asset type and industry. These productive capital stock estimates, together with
estimates of the stock of land and inventories, are used to derive the capital services
volume measure used in MFP calculations.
When an index of capital services is estimated, what is effectively calculated is a
weighted average movement of the various assets that make up the total productive
capital stock.1 Volume measures of productive capital stock can be interpreted as a
measure of the potential capital services that any given asset or group of assets can
contribute to the production process. Weights are based on asset-specific rental prices
(user costs). User costs represent the implied cost to the user of each type of capital.
The volume of capital services that flows from a given level of capital stock is assumed to
be proportional. The rate of utilisation of this capital is assumed to be constant over time.
This is reflected in the hypothetical example presented in figure 2. Assuming that the rate
of capacity utilisation is 90 percent, for any capital asset in the model the measured
capital stock utilised tracks along exactly parallel to the capital stock available.
In reality, this is unlikely for many capital assets (Hulten, 2001) and capacity utilisation
rates will tend to vary over time. As an economy experiences above-trend growth, firms
have to use their capital more intensively in the short term, since there are significant lead
times associated with making investment in new capital assets, and also in bringing such
investments online. This increased intensity could be achieved, for example, by having
factory workers work overtime (without changing the stock of capital), which would
increase the volume of output without a corresponding increase in the estimated volume
of input of capital services. Figure 3 presents a hypothetical example where actual capital
in use fluctuates according to a varying rate of utilisation. This implies that short-term
movements in capital input growth may not be calculated correctly if the rate of capacity
utilisation varies.
1
The productive capital stock is gross capital stock adjusted for the decline in its efficiency.
10
Adjusting productivity statistics for variable capacity utilisation: Working harder or hardly working?
Figure 2
2 Theoretical utilisation of capital, capital stock available and capital stock utilised
Hypothetical example: Constant utilisation of capital
Capital stock available and capital stock utilised
1978–2009
10
Capital volume
$(million)
Utilisation rate
1.0
8
0.8
6
0.6
4
0.4
2
0
Capital stock available
Capital stock utilised
0.2
Utilisation rate
78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 00 01 02 03 04 05 06 07 08 09
0.0
Year ended March
Source: Author's calculations.
Figure 3
3 Actual utilisation of capital, capital stock available and capital stock utilised
Hypothetical example: Variable utilisation of capital
Capital stock available and capital stock utilised
1978–2009
10
Capital volume
$(million)
Utilisation rate
1.0
8
0.8
6
0.6
4
0.4
2
0
Capital stock available
Capital stock utilised
Utilisation rate
78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 00 01 02 03 04 05 06 07 08 09
0.2
0.0
Year ended March
Source: Author's calculations.
Productivity measurement and interpretation
A measure of capital inputs is essential to calculating MFP. Statistics NZ’s method of
estimating productivity is based on the OECD’s (2001) guidelines.2 Calculating industry
productivity statistics begins by postulating a production function of the form:
(1)
𝑉𝑖 = 𝐴𝑖 (𝑡). 𝑓(𝐿𝑖 , 𝐾𝑖 (𝜃𝑖 ))
2
Further detail on the methodology Statistics NZ adopts in measuring productivity is found in
Productivity Statistics: Sources and methods.
11
Adjusting productivity statistics for variable capacity utilisation: Working harder or hardly working?
where 𝑉𝑖 = industry value added index
𝐿𝑖 = industry labour inputs
𝐾𝑖 = industry capital inputs, which is a function of 𝜃𝑖
𝜃𝑖 = rate of capacity utilisation of capital (assumed to be constant in official measures)
𝑓(𝐿𝑖 , 𝐾𝑖 (𝜃𝑖 )) = a production function of 𝐿𝑖 and 𝐾𝑖 that defines an expected level of output
for a specific industry
𝐴𝑖 (𝑡) = a parameter that captures disembodied technical shifts over time; that is, outward
shifts of the production function allowing output to increase with a given level of inputs
(known as MFP).
Given output, labour input, and capital input indexes, partial productivity measures can be
derived. Labour productivity for each industry is calculated as:
(2)
𝐿𝑃𝑖 = 𝑉𝑖 /𝐿𝑖
𝐿𝑃𝑖 is an index of labour productivity. This is a value-added index divided by a volume
index of labour inputs.
Similarly, capital productivity for each industry can be calculated as:
(3)
𝐾𝑃𝑖 = 𝑉𝑖 /𝐾𝑖 ((𝜃𝑖 )
𝐾𝑃𝑖 is an index of capital productivity. This is a value-added index divided by a volume
index of capital inputs. Holding 𝜃𝑖 constant, capital productivity can be calculated as:
𝐾𝑃𝑖 = 𝑉𝑖 /𝐾𝑖 .
Caution in interpreting the partial measures of productivity is recommended. For example,
labour productivity only partly measures 'true' labour productivity, in the sense of
capturing the personal capacities of workers or the intensity of their efforts. However,
labour productivity also reflects the change in capital available per worker and how
efficiently labour is combined with the other factors of production.
The parameter that represents disembodied technological change (or MFP) cannot be
observed directly. By rearranging the production function equation, it can be shown that
the technology parameter can be derived residually – as the difference between the
growth in an index of outputs and the growth in an index of inputs:
(4)
𝐴𝑖 (𝑡) = 𝑉𝑖 / 𝑓(𝐿𝑖 , 𝐾𝑖 (𝜃𝑖 ))
MFP growth can arise from, for example, advances in knowledge, improvements in
management, or production techniques. Certain assumptions must be met for MFP to be
a measure of disembodied technological change. The key assumptions are that the
production function must exhibit constant returns to scale, all inputs are included in scope
of the production function, and factor markets are perfectly competitive.
In practice, these conditions will not often be met and the resulting MFP residual needs to
be interpreted with some caution. Given the importance of technological progress as an
explanatory factor in economic growth, attention often focuses on the MFP measure as
though it is a measure of technological change. However, this is not often the case. When
interpreting MFP, the following should be noted.
•
Not all technological change translates into MFP growth. Embodied technological
change, such as advances in the quality of capital or improved human capital, will
be captured in the measured contributions of the inputs – provided they are
measured correctly (ie the volume input series includes quality change).
•
Observed MFP growth is not necessarily caused by technological change. Other
non-technology factors will be picked up by the residual, including economies of
scale, cyclical effects, inefficiencies, and measurement and model
misspecification errors (such as variable capacity utilisation).
12
3 Availability of data for adjusting for variable capacity
utilisation
Adjusting for a variable rate of capacity utilisation is possible for selected industries. This
section considers the key data sources for making these adjustments: the New Zealand
Institute of Economic Research’s (NZIER) capacity utilisation index for manufacturers and
builders; and NZIER’s capacity as a constraint series for merchants and services. 3
New Zealand Institute of Economic Research’s
capacity utilisation index
NZIER has conducted a comprehensive survey of business opinion – known as the
Quarterly Survey of Business Opinion (QSBO) – since 1961. This survey asks
respondent businesses about their output, costs and prices, and employment and
investment intentions. It also measures their perceptions of general business conditions.
The survey data are widely used as indicators for assessing various aspects of New
Zealand’s macro-economy.
One question in the survey addresses the intensity with which firms are using their plant
and equipment: “Excluding seasonal factors, by how much is it currently practicable for
you to increase your production from your existing plant and equipment without raising
unit costs?” Respondents can select one of five ranges: 0 percent, 1–5 percent, 6–10
percent, 11–20 percent, and over 20 percent. This question has remained unchanged
since the beginning of the survey.
The capacity utilisation, business opinion index (CUBO) is calculated from manufacturing
and building sector responses to this question. CUBO is calculated as a percentage – by
setting actual output equal to 100 and dividing by capacity output (100 plus the median
value of spare capacity). 4
Before physical constraints on production become binding, most firms start to experience
an increase in their average cost of production as output increases (assuming no change
in the level of plant and equipment used). For instance, higher average costs could arise
from the need to operate extra shifts, undertake additional plant maintenance, and so on.
This ‘economic capacity’ definition of capacity utilisation corresponds closely with that
used by NZIER (Hodgetts, 2004).
However, CUBO has limitations.
•
•
It is limited to manufacturers and builders.
There is no capital asset dimension available (although the question in the QSBO
refers to plant and equipment, it does not specify asset types).
CUBO is inherently cyclical in its behaviour, fluctuating over the growth cycle (see figure
4). 5 Over longer periods, CUBO may also be affected by structural changes in the
3
Occupancy rate data from Statistics NZ’s Accommodation Survey and Transpower’s transformer
capacity data were considered, but deemed impractical for this investigation. In the absence of data,
rates of capacity utilisation can be estimated using econometric methods (Shaikh & Moudud, 2004). A
further method of estimating capacity utilisation is to use appropriate survey data (Morin & Stevens,
2004). However, data used in this method are not available for New Zealand.
4
See Hodgetts (2004) for further detail on deriving the CUBO.
5
Quarterly data from NZIER’s capacity utilisation index was converted into an annual series by taking
March-year averages.
13
Adjusting productivity statistics for variable capacity utilisation: Working harder or hardly working?
economy. Changes in productivity, working patterns, cost structures, or technology could
all potentially affect the average level of CUBO prevailing over time. Despite these
limitations, there is still validity in investigating how adjusting the capital services series
would affect the resulting MFP estimates – by adjusting the input productive capital stock
series by the CUBO ratio, to approximate the ideal ‛machine hour’ measure of capital.
Figure 4
4 Capacity utilisation and GDP, annual percentage change
Capacity utilisation and GDP
Annual percentage change
1978–2009
8
Percent
6
Capacity utilisation
Measured sector GDP
4
2
0
-2
-4
79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 00 01 02 03 04 05 06 07 08 09
Year ended March
Source: Author's calculations using Statistics New Zealand and NZIER data
Adjusting for capacity utilisation will have more of an impact if the average rate of
utilisation has shifted over time. While there appears to have been an increasing trend in
capacity utilisation, New Zealand may be unique in this regard. Etter, Graff, and Muller
(2008) found capacity utilisation in OECD countries has, on average, been trending
downwards since 1970. This implies the impact of adjustment in New Zealand may have
a different effect to that in other countries. Comparing capacity utilisation rates across
growth cycles used by Statistics NZ, it can be seen that the average capacity utilisation
rate has been only mildly variable, apart from an upswing since 2000 (see table 1).
Capacity utilisation has also stabilised, with the standard deviation lessening slightly over
time. While there is minimal variation in capacity utilisation the long term, relatively high
rates of change can be observed in the short term. The impact on MFP may therefore be
more pronounced in the short-term.
Table 1
1 Capacity utilisation across growth cycles
Capacity utilisation across growth cycles
Growth cycle
Average rate of capacity utilisation
Standard deviation
1982–85
0.88
0.021
1985–90
0.87
0.017
1990–97
0.88
0.025
1997–2000
0.88
0.012
2000–06
0.91
0.010
Source: Authors’ calculations using New Zealand Institute of Economic Research data
14
Adjusting productivity statistics for variable capacity utilisation: Working harder or hardly working?
Separate capacity utilisation series for manufacturing and construction are available.
Figure 5 highlights the series for manufacturing and construction, and the combined
CUBO. The manufacturing series follows that of the combined series closely, due to the
large weight of manufacturers. The construction series exhibits greater volatility than the
manufacturing series. All else being equal, this implies that capacity adjustment has a
greater impact on construction. As with the combined series, the level of average
utilisation has increased for both the manufacturing and construction series.
Figure 5
5 Capacity utilisation rates, combined and industry specific CUBO series
Capacity utilisation rates
Combined and industry specifc CUBO series
0.95
1978–2010
Utilisation rate
Manufacturers and builders
Manufacturers
Builders
0.90
0.85
0.80
0.75
0.70
78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 00 01 02 03 04 05 06 07 08 09 10
Year ended March
Source: Author's calculations using NZIER data
Figure 6 shows the movements in capacity utilisation, capital input, and output in the
manufacturing industry. There is a relatively strong relationship between output growth
and CUBO growth, which highlights that adjusting manufacturing’s capital services with
CUBO is appropriate.
Figure 6
6 Manufacturing, capacity utilisation, capital input, and output growth
Manufacturing
Capacity utilisation, capital input, and output growth
15
Percent
1978–2009
Capacity utilisation
Capital input
Output
10
5
0
-5
-10
79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 00 01 02 03 04 05 06 07 08 09
Year ended March
Source: Author's calculations using Statistics New Zealand and NZIER data
Capacity utilisation figures for the construction industry also strongly correlate with its
output growth from 1978, highlighting strong procyclicality within the capacity utilisation
measure. The relationship between capacity utilisation and output is strong, although the
15
Adjusting productivity statistics for variable capacity utilisation: Working harder or hardly working?
relationship is weakened by two outliers. 6 These observations highlight the value in using
CUBO (construction) for adjustment in this industry. Figure 7 highlights the volatility in
output within this industry, and the positive relationship between output and capacity
utilisation.
Figure 7
7 Output growth and capacity utilisation growth, construction
Output growth and capacity utilisation growth
Construction
1979–2009
Output growth
20
15
10
5
-10
-5
0
-5
0
5
10
15
20
-10
-15
-20
Capacity utilisation
Source: Author's calculations using Statistics New Zealand and NZIER data
New Zealand Institute of Economic Research’s
capacity as a constraint series
Capacity utilisation is mainly discussed with reference to manufacturing and construction,
but variable utilisation rates may also be present for service industries. Indeed, if the
impact of capacity utilisation on the measured economy is to be assessed, then data
sources for other industries need to be examined.
The QSBO is again useful for addressing variable capacity utilisation by asking “what
single factor, if any, is most limiting your ability to increase turnover?” Respondents can
choose: ‘demand’, ‘supplies’, ‘finance’, ‘labour’, ‘capacity’, or ‘other’. The capacity as a
constraint (CAAC) series is derived from responses to this question. It is available for
manufacturers and builders, merchants (wholesale trade and retail trade), services, and
the total economy. Industry-specific CAAC series can be derived where appropriate.
The series is based on the proportion of respondents reporting that capacity is
constraining. As such, it is not a direct measure of capacity utilisation. An index of
capacity utilisation from the CAAC series was created by setting the 1977 capacity
utilisation point to 0.8, and then indexing with the difference in CAAC. 7 Although the
choice of 0.8 for the start year may seem arbitrary, a number of studies (eg Shaikh &
Moudud, 2004, or Etter et al, 2008), and even the CUBO, suggest this is a common level
for capacity utilisation. The choice of 0.8 as a base year level for the utilisation rate
assumes that capacity utilisation in service industries is at roughly the same level as in
manufacturing and construction. However, the level of utilisation in accommodation,
6
These outliers (in 1980 and 1993) potentially suggest that the increase in capacity utilisation is
constraining output. This may have led to sharply increasing costs, foregone investment in new (and
potentially relatively cheaper) capital.
7
Differences rather than percentage changes were used in the indexation as the CAAC is based on
percentage of respondents. An analysis of CUBO and CAAC shows that the differences in CAAC and
percentage changes in CUBO are strongly related.
16
Adjusting productivity statistics for variable capacity utilisation: Working harder or hardly working?
cafes, and restaurants may be lower (around 0.5), as indicated by occupancy rate data. A
sensitivity test for the level of utilisation showed the general impact of capacity
adjustment is consistent across reasonable levels. 8
The CAAC series is applied to capital stock data in the same manner as the CUBO (for
consistency) despite the CAAC referring to capacity in a broader sense than just capital
and may pick up capacity issues relating to labour. As the OECD (2003) state:
Some respondents, however, will take account of other factors such as access to
financial capital and, particularly, the supply of labour. Again this should not affect the
validity of the results so far as changes over time are concerned provided that
respondent behaviour is stable. However, survey data on the actual levels of capacity
utilisation will represent some unknown mixture of capital and labour utilisation.
The service sector in the CAAC series does not cover all service industries that are part
of the measured sector. Wholesale trade; retail trade; finance and insurance; business
services; accommodation, cafes, and restaurants; and transport and storage are
adequately covered by QSBO. Industry-specific CAAC utilisation series can therefore be
applied to the capital stock data for these industries. While some activity in property
services, and personal and other community services is covered in QSBO, their coverage
is only partial and therefore not deemed representative.
In addition, growth rates in CAAC for services align with the output growth of service
industries to varying degrees. 9 The output and capacity utilisation series are strongly
related for both transport and storage, and business services. The industry-specific
CAAC series show weaker relationships for accommodation, cafes, and restaurants, and
retail trade than CAAC for services. Although it may appear that the industry-specific
measures are less applicable, it could reflect substantially large movements from belowoptimal capacity output to above (or vice versa).
Relationship between CUBO and CAAC
The capacity utilisation measures are relevant to different industries to different degrees,
but they also correlate across industries (see table 2). For example, capacity utilisation
for accommodation, cafes, and restaurants is weakly related to that in all other industries
(except business services). In contrast, capacity utilisation in manufacturing is much more
correlated with that in other industries (except accommodation, cafes, and restaurants;
and finance and insurance). This highlights the importance of applying industry-specific
capacity utilisation data, rather than generic sector series to industries.
8
As the level of utilisation may not reflect the true level of utilisation, this measure may be limited when
considering capital and MFP levels.
9
The measured-sector service sector includes (from 1978): wholesale trade; retail trade;
accommodation, cafes and restaurants; transport and storage; communication services; finance and
insurance; and cultural and recreational services. Property and business services and personal and
other community services are included from 1996.
17
Adjusting productivity statistics for variable capacity utilisation – Working harder or hardly working?
Table 2
2 Correlation matrix for industry-level capacity utilisation measures
Correlation matrix for industry-level capacity utilisation measures
CUBO
CUBO
Manufacturing
Construction
CAAC
Wholesale
trade
Retail trade
Accommodation,
cafes, &
restaurants
Transport &
storage
Finance &
insurance
Business
services
1
Manfacturing
0.98
1
Construction
0.80
0.69
1
CAAC
0.78
0.74
0.55
1
Wholesale trade
0.67
0.65
0.49
0.68
1
Retail trade
0.62
0.56
0.51
0.81
0.71
1
Accommodation,
cafes, & restaurants
0.28
0.29
-0.01
0.50
0.33
0.33
1
Transport & storage
0.58
0.60
0.29
0.75
0.58
0.70
0.21
1
Finance & insurance
0.37
0.30
0.44
0.63
0.36
0.62
0.28
0.41
1
Business services
0.56
0.56
0.25
0.84
0.46
0.50
0.52
0.54
0.38
18
1
4 Applying capacity utilisation measures to capital
assets
This section considers the question ‘which assets should be adjusted?’ The answer
depends on the data and the nature of the assets.
Variable capacity utilisation rates are applied at the asset level. NZIER’s CUBO index
specifically relates to plant, equipment, and machinery assets. This does not preclude
other assets being adjusted. If the plant is not being fully utilised then arguably neither is
the other equipment in the factory. If plant or equipment are not fully utilised then demand
is likely to be low – there is likely to be less use of transport equipment to deliver goods or
of information technology to process orders.
The utilisation rate was applied to all assets other than land, buildings, and inventories –
they enable production whether or not they are in use. For example, a building (and the
land it is on) is a necessary requirement for any level of production. For these assets, it is
assumed that the utilisation rate is proportional to the volume of productive capital stock.
This is not an unreasonable assumption.
Although the CAAC data is not related to any specific asset types, the utilisation rate was
applied to the same set of assets as above.
The capital input data were adjusted by multiplying the underlying capital stock data for
the relevant assets and industries by the utilisation rate. The components of the user cost
equation were not adjusted. 10 This is consistent with the argument from Berndt and Fuss
(1986) in that the ex post user cost of capital does not need to be adjusted.
10
User costs are a function of the price index of capital assets, the real rate of return (set at 4 percent),
the rate of depreciation, and the average non-income tax rate on production. See Statistics NZ (2012)
for more detail.
19
5 Empirical analysis
Productivity estimates for the measured sector, and for each industry adjusted using
available data on variable utilisation, were derived under the assumption of a constant
rate of utilisation (the base case), and using a variable utilisation rate (capacity
adjusted). 11 The results for the measured sector are discussed here, and the impact of
adjusting for variable capacity utilisation on specific industries is in the appendix.
Industries adjusted by CUBO and CAAC data covered 56.3 percent of the total economy
(which is 70.1 percent of the measured sector) in terms of current price gross domestic
product in 2007. 12 The capacity utilisation measures, and industries they can be applied
to, are noted in table 3. All capacity utilisation series start in 1977.
Table 3
3 Summary of capacity adjusted measures and industries
Summary of capacity adjusted measures and industries
Capacity utilisation
measure
Industries adjusted
CUBO – industry specific
Manufacturing; construction
CAAC – industry specific
Wholesale trade; retail trade; accommodation, cafes, and
restaurants; transport and storage; business services;
finance and insurance
Impact on measured sector input
Capacity adjustment led to a decrease in capital input growth for 1978–2010.
Interestingly, the main points of divergence between the adjusted and base-case series
for capital input occur at the same time as economic recessions (see figure 8).
11
Output data are consistent with those published in the September 2011 quarter GDP release.
12
Industries included in the measured sector but not adjusted are: agriculture, forestry, and fishing;
mining; electricity, gas, and water supply; communication services; property services; personal and other
community services; and cultural and recreational services. Government administration and defence,
health, and education are not included in the measured sector.
20
Adjusting productivity statistics for variable capacity utilisation: Working harder or hardly working?
Figure 8
8 Measured sector capital services, with and without capacity adjustments, annual percentage change
Measured sector capital services
With and without capacity adjustments, annual percentage change
8
1979–2010
Percent
Base case
6
Capacity adjusted
4
2
0
-2
79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 00 01 02 03 04 05 06 07 08 09 10
Year ended March
Source: Author's calculations using Statistics New Zealand and NZIER data
Figure 9
9 Measured sector capital productivity, with and without capacity adjustments, annual percentage change
Measured sector capital productivity
With and without capacity adjustments, annual percentage change
8
6
Percent
Base case
1979–2010
Capacity adjusted
4
2
0
-2
-4
-6
79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 00 01 02 03 04 05 06 07 08 09 10
Year ended March
Source: Author's calculations using Statistics New Zealand and NZIER data
The impact of capacity adjustment (CUBO and CAAC) also alters the depiction of the
economy during the recent economic downturn. Statistics NZ’s (2010) Productivity
Statistics: 1978–2009 information release states:
Although capital productivity declined sharply in 2009, it is possible utilisation of
capital was at a low point. …Under conditions where utilisation of capital is lower
than average, growth in capital inputs may be artificially high and therefore growth in
capital productivity may be artificially low.
This scenario is observable in figures 8 and 9. After adjusting for variable capacity
utilisation, capital input growth in 2009 was 1.1 percent, which contrasts with growth of
2.8 percent in the base case. This led to a smaller fall in capital productivity growth.
As capacity adjustment has led to a decrease in capital input growth, in total input growth
has also decreased (figure 10). The impact on total inputs is less than that for capital
inputs as it is weighted by the movements in both labour and capital input.
21
Adjusting productivity statistics for variable capacity utilisation: Working harder or hardly working?
Figure 10
10 Measured sector total inputs, with and without capacity adjustments, annual percentage change
Measured sector total inputs
With and without capacity adjustments, annual percentage change
6
1979–2010
Percent
Base case
Capacity adjusted
4
2
0
-2
-4
79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 00 01 02 03 04 05 06 07 08 09 10
Year ended March
Source: Author's calculations using Statistics New Zealand and NZIER data
Impact on measured sector MFP
As total inputs with a variable rate of capacity utilisation grew more slowly than under the
base case, measured sector MFP was greater after adjustment (see figure 11).
While the series occasionally diverge, they generally converge over the longer-term.
These observations concur with the expectation that while capacity adjustment may not
have much impact in the long term, it improves MFP estimates between peak years.
Figure 11
11 Measured sector MFP, with and without capacity adjustments, annual percentage change
Measured sector MFP
With and without capacity adjustments, annual percentage change
1979–2010
6
4
Percent
Base case
Capacity adjusted
2
0
-2
-4
79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 00 01 02 03 04 05 06 07 08 09 10
Year ended March
Source: Author's calculations using Statistics New Zealand and NZIER data
Adjusting for variable capacity utilisation had the greatest effects on MFP in construction,
wholesale trade, and finance and insurance. That said, in none of these industries did
capacity adjustment dramatically change the long-term MFP trend. The impact is
illustrated in the appendix.
22
6 Summary
This paper has assessed the impact of adjusting official productivity statistics with
independently derived measures of capacity utilisation of capital, to best reflect
measured-economy and industry-level productivity growth.
Adjustment can potentially improve the reliability of estimates, especially in turbulent
periods in the economy. The empirical analysis in this paper indicates that capacity
adjustment has minimal effect on long-term MFP growth patterns, and the effect is mainly
seen on year-on-year movements. Therefore, while the assumption of a constant rate of
capacity utilisation may be seen as theoretically restrictive, in the long term there is little
impact in practice.
23
References
Basu, S (1996). Procyclical productivity: Increasing returns or cyclical utilization? The
Quarterly Journal of Economics, 3(3), 719–751. Available from www.jstor.org.
Basu, S, & Fernald, J (2000). Why is productivity procyclical? Why do we care? (Federal
Reserve Bank of Chicago working paper series 00-11). Available from
www.chicagofed.org.
Berndt, E, & Morrison, C (1981). Capacity utilization measures: Underlying economic
theory and an alternative approach. American Economic Review, 71(2), 48–52. Available
from www.jstor.org.
Berndt, E, & Fuss, M. (1986). Productivity measurement with adjustments for variations in
capacity utilization, and other forms of temporary equilibrium. Journal of Econometrics,
33; 7–29.
Etter, R, Graff, M, & Muller, J (2008, October). Is ‘normal’ capacity utilisation constant
over time? Analyses with macro and micro data from business tendency surveys. Paper
presented at 29th CIRET Conference 2008, Santiago de Chile.
Hornstein, A (2002). Towards a theory of capacity utilization: Shiftwork and the workweek
of capital. Federal Reserve Bank of Richmond Economic Quarterly, 88/2 Spring 2002.
Hodgetts, B (2004). NZIER’s capacity utilisation index. Reserve Bank of New Zealand
Bulletin, 67(3), 19–26. Available from www.rbnz.govt.nz.
Hulten, C (2001). Total factor productivity: A short biography. In C Hulten, E Dean, & M
Harper (Eds), New developments in productivity analysis (1–54). Chicago, Illinois:
University of Chicago Press.
Morin, M, & Stevens, J (2004). Estimating Capacity Utilization from Survey Data, Finance
and Economics. (Discussion series) Divisions of Research & Statistics and Monetary
Affairs Federal Reserve Board, Washington, D.C.
Nelson, R (1989). On the measurement of capacity utilization. The Journal of Industrial
Economics, 37(3), 273–286.
Organisation of Economic Co-operation and Development (2001). Measuring productivity:
Measurement of aggregate and industry-level productivity growth. Paris: OECD. Available
from www.oecd.org.
Organisation of Economic Co-operation and Development (2003). Business tendency
surveys: A handbook. Paris: OECD. Available from www.oecd.org.
Shaikh, A, & Moudud, J (2004). Measuring capacity utilisation in OECD countries: a
cointegration method (Levy Economics Institute economics working paper archive
number wp_415). New York: Levi Economics Institute. Available from
www.levyinstitute.org.
Statistics New Zealand (2010). Productivity Statistics: 1978–2009. Available from
www.stats.govt.nz.
Statistics New Zealand (2012). Productivity statistics: Sources and methods. Available
from www.stats.govt.nz.
24
Appendix: Industry effects
Manufacturing
Manufacturing: MFP with and without capacity adjustment, indexes
Manufacturing
MFP with and without capacity adjustment, indexes
1300
1978–2009
Index
MFP (capacity adjusted)
MFP (base case)
1200
1100
1000
900
78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 00 01 02 03 04 05 06 07 08 09
Year ended March
Source: Author's calculations using Statistics New Zealand and NZIER data
Manufacturing: MFP with and without capacity adjustment, annual percentage change
Manufacturing
MFP with and without capacity adjustment, annual percentage change
8
6
1979–2009
Percent
MFP (capacity adjusted)
MFP (base case)
4
2
0
-2
-4
-6
-8
79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 00 01 02 03 04 05 06 07 08 09
Year ended March
Source: Author's calculations using Statistics New Zealand and NZIER data
25
Adjusting productivity statistics for variable capacity utilisation – Working harder or hardly working?
Construction
Construction: MFP with and without capacity adjustment, indexes
Construction
MFP with and without capacity adjustment, indexes
1300
1978–2009
Index
MFP (capacity adjusted)
MFP (base case)
1200
1100
1000
900
800
78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 00 01 02 03 04 05 06 07 08 09
Year ended March
Source: Author's calculations using Statistics New Zealand and NZIER data
Construction: MFP with and without capacity adjustment, annual percentage change
Construction
MFP with and without capacity adjustment, annual percentage change
10
1979–2009
Percent
5
0
-5
-10
-15
MFP (capacity adjusted)
MFP (base case)
79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 00 01 02 03 04 05 06 07 08 09
Year ended March
Source: Author's calculations using Statistics New Zealand and NZIER data
26
Adjusting productivity statistics for variable capacity utilisation – Working harder or hardly working?
Wholesale trade
Wholesale trade: MFP with and without capacity adjustment, indexes
Wholesale trade
MFP with and without capacity adjustment, indexes
1200
1978–2009
Index
MFP (capacity adjusted)
MFP (base case)
1100
1000
900
800
78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 00 01 02 03 04 05 06 07 08 09
Year ended March
Source: Author's calculations using Statistics New Zealand data
Wholesale trade: MFP with and without capacity adjustment, annual percentage change
Wholesale trade
MFP with and without capacity adjustment, annual percentage change
15
10
1979–2009
Percent
MFP (capacity adjusted)
MFP (base case)
5
0
-5
-10
-15
79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 00 01 02 03 04 05 06 07 08 09
Year ended March
Source: Author's calculations using Statistics New Zealand data
27
Adjusting productivity statistics for variable capacity utilisation – Working harder or hardly working?
Retail trade
Retail trade: MFP with and without capacity adjustment, indexes
Retail trade
MFP with and without capacity adjustment, indexes
1100
1978–2009
Index
MFP (capacity adjusted)
MFP (base case)
1000
900
800
78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 00 01 02 03 04 05 06 07 08 09
Year ended March
Source: Author's calculations using Statistics New Zealand data
Retail trade: MFP with and without capacity adjustment, annual percentage change
Retail trade
MFP with and without capacity adjustment, annual percentage change
1979–2009
8
6
Percent
MFP (capacity adjusted)
MFP (base case)
4
2
0
-2
-4
-6
79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 00 01 02 03 04 05 06 07 08 09
Year ended March
Source: Author's calculations using Statistics New Zealand data
28
Adjusting productivity statistics for variable capacity utilisation – Working harder or hardly working?
Accommodation, cafes, and restaurants
Accommodation, cafes, and restaurants: MFP with and without capacity adjustment, indexes
Accommodation, cafes, and restaurants
MFP with and without capacity adjustment, indexes
1100
1978–2009
Index
MFP (capacity adjusted)
1000
900
800
700
600
78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 00 01 02 03 04 05 06 07 08 09
Year ended March
Source: Author's calculations using Statistics New Zealand data
Accommodation, cafes, and restaurants: MFP with and without capacity adjustment, annual percentage change
Accommodation, cafes, and restaurants
MFP with and without capacity adjustment, annual percentage change
5
1979–2009
Percent
0
-5
-10
-15
MFP (capacity adjusted)
MFP (base case)
79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 00 01 02 03 04 05 06 07 08 09
Year ended March
Source: Author's calculations using Statistics New Zealand data
29
Adjusting productivity statistics for variable capacity utilisation – Working harder or hardly working?
Transport and storage
Transport and storage: MFP with and without capacity adjustment, indexes
Transport and storage
3000
Index
MFP with and without capacity adjustment, indexes
1978–2009
MFP (base case)
MFP (capacity adjusted)
2500
2000
1500
1000
500
78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 00 01 02 03 04 05 06 07 08 09
Year ended March
Source: Author's calculations using Statistics New Zealand data
Transport and storage: MFP with and without capacity adjustment, annual percentage change
Transport and storage
MFP with and without capacity adjustment, annual percentage change
1979–2009
20
Percent
MFP (base case)
15
MFP (capacity adjusted)
10
5
0
-5
-10
79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 00 01 02 03 04 05 06 07 08 09
Year ended March
Source: Author's calculations using Statistics New Zealand data
30
Adjusting productivity statistics for variable capacity utilisation – Working harder or hardly working?
Finance and insurance
Finance and insurance: MFP with and without capacity adjustment, indexes
Finance and insurance
MFP with and without capacity adjustment, indexes
1800
1978–2009
Index
MFP (capacity adjusted)
MFP (base case)
1600
1400
1200
1000
800
78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 00 01 02 03 04 05 06 07 08 09
Year ended March
Source: Author's calculations using Statistics New Zealand data
Finance and insurance: MFP with and without capacity adjustment, annual percentage change
Finance and insurance
MFP with and without capacity adjustment, annual percentage change
1979–2009
10
Percent
MFP (capacity adjusted)
MFP (base case)
5
0
-5
-10
79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 00 01 02 03 04 05 06 07 08 09
Year ended March
Source: Author's calculations using Statistics New Zealand data
31
Adjusting productivity statistics for variable capacity utilisation – Working harder or hardly working?
Business services
Business services: MFP with and without capacity adjustment, indexes
Business services
MFP with and without capacity adjustment, indexes
1050
1996–2009
Index
MFP (capacity adjusted)
MFP (base case)
1000
950
900
850
96
97
98
99
00
01
02
03
04
Year ended March
05
06
07
08
09
Source: Author's calculations using Statistics New Zealand data
Business services: MFP with and without capacity adjustment, annual percentage change
Business services
MFP with and without capacity adjustment, annual percentage change
1997–2009
4
Percent
MFP (capacity adjusted)
MFP (base case)
2
0
-2
-4
-6
97
98
99
00
01
02
03
04
Year ended March
Source: Author's calculations using Statistics New Zealand data
32
05
06
07
08
09