Appendix - Sources and Methods

A1   Principal data sources

A1.1  Output measures

For the UK National Accounts data on current price gross output and value added were derived from the Office of National Statistics (ONS) Blue Book (2006) and ONS Supply-Use Tables 1995-2004. Real value added series were then derived using chain-linked volume indices published in the Blue Book and unpublished industry-specific output deflators supplied by ONS.

For New Zealand data on current price gross output and value added were derived from the National Accounts (Revised): Year ended March 2005. The conversion of March year data to a calendar year basis was far from straightforward. Statistics NZ supplied quarterly data on gross domestic product by industry in the form of a seasonally-adjusted chain-volume series expressed in 1995-96 prices (ie, in prices relating to the 12 months ending March 31, 1996) together with a quarterly Producers Price Index - Output series disaggregated by industry. These two data series were used to generate an industry-level current price value added index on a calendar year basis. However, the problem still arose of how to estimate a set of current-price starting values for this series. In the event current price value added totals for the calendar year 1995 were assumed to consist of 25% of current price value added for the 12 months ending March 1995 plus 75% of current price value added for the 12 months ending March 1996. The precision of our productivity comparisons could therefore be improved if quarterly data on current price value added disaggregated by industry could be made available.

In both countries output deflators for aggregate market sectors were calculated using a Tornqvist index formula with sector-level deflators weighted by each sector’s average share of total current price value added in adjacent years.

A1.2   Labour input measures

For UK employment ONS Blue Book totals at broad industry level were taken as control totals and then disaggregated to more detailed sector level using data on employment shares from the Annual Business Inquiry. Industry-level data on average annual hours worked per person engaged (including unpaid overtime) in the UK were derived from analysis of Labour Force Survey data.

For New Zealand the most reliable data series on labour inputs, described in Statistics New Zealand (2006b), shows total hours paid in a reference week in the middle of each quarter, including the self-employed as well as employees. Estimates of total annual hours worked by industry were obtained by summing the four weekly hours paid figures for each year, multiplying by 13 and then making a further adjustment to an hours worked basis, using an aggregate market sectors ratio of hours worked to hours paid derived from NZ Household Labour Force Survey (HLFS) data. This procedure was carried out on the advice of Statistics NZ because of concerns about lack of robustness at industry level in the HLFS hours worked series.

Remaining concerns about the comparability of labour input data concern the treatment of annual leave and other forms of absence from work in each country. The New Zealand hours worked series is based on responses collected over 52 weeks of the year and should therefore capture all forms of absence from work (paid or unpaid). For the UK the Labour Force Survey data on actual hours worked by survey respondents refer to a specified reference week which is usually the week prior to each interview. In general, LFS estimates are believed to take reasonable account of reduced working hours due to annual leave and statutory holidays and LFS methodological notes refer to procedures for some interviews falling during the Christmas/New Year period. Hence, it is likely that the UK estimates take much the same account of annual leave and other absences from work as do the New Zealand estimates.

A1.3   Labour share of value added

For the UK estimates of the labour share of value added at industry level were derived from Annual Business Inquiry data on employee compensation and value added, with an upward adjustment to take account of self-employed persons based on estimates of the ratio of self-employed to employees derived from Labour Force Surveys.

For New Zealand the labour share of value added at industry level was estimated in a similar manner using National Accounts data on employee compensation and value added at industry level, with an upward adjustment to take account of self-employed persons based on estimates of the ratio of self-employed to employees in the SNZ hours paid labour volume series described above.

For both countries we assume that self-employed hourly earnings are 70% of average hourly wages for employees. This procedure follows an approach suggested in O’Mahony and van Ark (2003) in the light of US evidence of generally lower compensation for self-employed persons compared to employees.

A1.4   Capital stocks

Capital stocks series were constructed using a perpetual inventory method that cumulates constant price investments and deducts the value of depreciated assets. Capital investment data at industry level were provided by the UK Office for National Statistics and Statistics New Zealand. In order to derive comparable estimates of productive capital stocks in New Zealand and the UK, common sector-specific depreciation rates (based on US estimates) were applied to the investment data in each country. Five asset types were distinguished: structures (non-residential buildings and other construction), computers, other plant and machinery, vehicles and intangibles (defined in the UK as consisting of patents, mineral exploration, artistic originals and the value of computer software). Investment data in national currencies were converted to US$ using OECD PPPs for investment goods by asset type. Finally, starting values for capital stocks were required in order to implement the perpetual inventory formula. In the UK starting values were set in 1948 by raising investment for that year by a factor equal to 0.5* (1/d_j) where d_j denotes the depreciation rate for asset type j.[14] In New Zealand the starting year for applying this formula ranged from 1859 for buildings to 1964 for computers.

Thus letting c denote types of capital, with I denoting investment and d the (geometric) depreciation rate, capital stocks were measured as:

The growth in aggregate capital was then calculated using a Tornqvist index formula, with weights equal to the share of each asset type in the total value of capital.

The assumption of geometric depreciation rates has the advantage that it is easy to implement. Its main disadvantage is that assets are depreciated rapidly at the beginning of the asset’s life but depreciation then tails off subsequently. This assumption is more reasonable for assets where technological change is rapid than it is for assets such as structures.

Estimated productive capital stocks were initially supplied for New Zealand in constant price 1995-96 NZ$ and for the UK in 2002 constant price £ sterling. Both these series were converted to constant price 1999 US$ using 1999 OECD PPPs for non-ICT investment goods by asset type along with deflators based on movements in investment goods producer price indices in the UK, US and New Zealand. For computers and software (assumed to be representative of intangibles), US ICT capital stock deflators were used, obtained from http://www.csls.ca/data/ict.asp