The Scottish Government

« Previous | Contents | Next »

Listen

ANALYSIS OF HISTORICAL CONSTRUCTION COST MOVEMENTS IN SCOTTISH SOCIAL HOUSING FINAL REPORT

2.0 CAPITAL COSTS

2.1 General

Under this heading, capital costs are distinguished from tender prices as being a measure of the movement of labour and materials costs associated with component replacement and improvement of social housing. These are notional cost movements and exclude the market place elements of profit and overheads, labour bonuses and materials discounts.

Building input cost index series nominally measure changes in costs incurred by contractors. The input cost index series available in the UK are factor cost indices, with cost levels being derived from a weighted average sample of labour and materials.

Building input cost indices illustrate the background cost pressure on contractors but have no component related to market conditions. Levels of overhead and profit, wages for non-direct labour or the effects of changes in productivity are not accounted for in an input cost index.

The data for the trends examined have been sourced from:

• Department of Trade and Industry (DTI)
• Office for National Statistics (ONS)
• Building Cost Information Service (BCIS)

2.2 Department of Trade and Industry

2.2.1 Housing and non-housing

The Department of Trade and Industry publish a series of Resource Cost Indices for Building Non-Housing and House Building, which are published in a quarterly document entitled ' Quarterly Building Price and Cost Indices'. The indices are prepared within the Construction Market Intelligence Division of the Construction Sponsorship Directorate of the Department of Trade and Industry. The indices are applicable on a UK wide basis and are compiled by application of the Price Adjustment Formulae for Building and Specialist Engineering Works (Series 3) to cost models. The detailed weightings of the Cost Models are not publicly available and DLE have been unable to secure access to these models.

The Price Adjustment Formulae for Building and Specialist Engineering Works indices originated in 1970 and have been refined and have progressively increased in number since then. The indices were devised as a simple way of calculating and reimbursing fluctuations on traditional building contracts with the agreement and participation of all sides of the industry.

The current series of indices have been running since 1990 and consist of 60 Building Work categories plus separate indices for scaffolding, plant, labour, electrical, mechanical, lift, catering and structural steelwork installations. In addition 15 index series are compiled for civil engineering work. A detailed description and build up of all the indices, as well as its use in calculating fluctuations payments, is given in ' Price Adjustment Formulae for Construction Contracts: Users Guide'.

The Labour components of the indices are based on the calculation of 'all-in' hourly rates of pay for operatives based on national wage agreements.

The Materials components of the indices are generally Producer Price Indices prepared by the Office for National Statistics.

The two principal input cost indices published by the DTI are:

Resource Cost Index of House Building - HOCOS

Resource Cost Index of Building Non-Housing - NOCOS

The relevance of the analysis of the HOCOS and NOCOS indices is the opportunity to compare trends between the housing and non-housing sectors, which in turn will determine the extent to which findings from non-housing specific indices and other data can be extrapolated to represent patterns of cost in the housing sector.

The HOCOS and NOCOS index series are further broken down into input/cost indices representing the costs associated with the following components:-

• General building works, mechanical and electrical services
• Labour and materials (derived indices only)

The breakdown of the index into these more detailed cost indices facilitates a greater understanding of the relative significance of these component costs.

Chart 2.1 plots, since 1985, the real movement of the Building Non-Housing (NOCOS) index with the movement of the House Building (HOCOS) index, both deflated to real terms by division by the GDP Deflator. The purpose is to determine whether there has been any significant differential movement between the housing and non-housing models.

The margin observable in the chart that exists between the housing and non-housing index series opened up between 1985 and 1986, since when the two series have moved largely in parallel. The reason for the differential movement between 1985 and 1986 (nominal increase of 4.9% for NOCOS but only 1.6% for HOCOS) is not readily apparent without access to the models involved.

The margin is considerably more than might have been expected from examination of individual inputs. The principal inputs to the House Building index would be expected to be brickwork and blockwork, concrete, carpentry, joinery, central heating and electrical installations. Carpentry increased by 1.2% in 1986 and hardwood joinery by only 0.4% but brickwork and blockwork, softwood joinery, boilers, pumps and radiators and plumbing and electrical labour all rose by more than 4%. However, many items more associated with non-housing work, such as aluminium windows, steel pipes, screeds, tiling, suspended ceilings, heating and ventilating labour and structural steelwork increased by 5% or more.

Chart 2.1 Housing and Non-housing Resource Cost Indices

Comparing housing and non-housing work, average real (i.e. inflation adjusted) cost trends across the UK as a whole have been broadly similar over 17 years between 1985 and 2002. Housing costs increased by 3.2% over the whole period (average 0.2% per annum), while non-housing costs increased by 5.1% over the whole period (average 0.3% per annum). Both index series dipped below general inflationary trends between 1990 and 1992, followed by a reversal of the trend between 1992 and 1995.

Over the first 7 years of the period (between 1985 and 1992), real UK housing capital costs fell by 7%. This was a larger fall than for non-housing costs (which fell by 3%). Since then cost increases have outstripped underlying inflation, with real costs rising by 11% to 2002 for housing and 8% for non-housing. The index series started to rise at a faster rate from 1999. This was a result of wage awards in the building industry exceeding wage increases in industry generally, as construction industry workload exceeded GDP growth.

These trends confirm the cyclical nature of construction costs in the UK.

Chart 2.1 indicates that there is no significant difference in the input costs associated with housing and non-housing and that it would be reasonable to apply non-housing based data to this analysis. The graph also indicates that, following the period of reducing costs, factors are currently driving costs ahead of underlying inflation and these factors are examined in greater detail below.

2.2.2 Building and engineering services

The DTI Resource Cost Indices are broken down to show the individual cost movement of Building, Heating and Ventilating and Electrical works. More detailed analysis was carried out to examine any differing trends in these sectors. In summary, over the whole period, mechanical and electrical services works have increased slightly more than building work, but building work has risen at a slightly faster rate over the last four years.

Chart 2.2 plots each of these components within the Resource Cost Index of House Building (HOCOS), deflated to portray real cost trends.

The Mechanical and Electrical components of the index have risen more than the Building component over the study period. Neither of the engineering index series experienced the fall relative to underlying inflation that occurred in the building sector between 1985-86 nor 1990-92. Electrical work costs have risen by the greatest amount in real terms over the study period, rising 10.2% overall (or an average 0.6% per annum). Since 1997, costs of mechanical work have increased in real terms by just 0.1% per annum, while building work has increased at a faster rate than the engineering components as a result of higher annual wage awards.

Cost movements affecting the Mechanical, Electrical and Building components of the Resource Cost Index of Building: Non-Housing (NOCOS), were also examined and were found to be broadly similar to the results for HOCOS above.

Chart 2.2 Mechanical, Electrical and Building Components of HOCOS

2.2.3 Labour, Plant and Materials

The Resource Cost Indices are also sub-divided into Labour & Plant and Materials indices. These have been plotted on Chart 2.3 to determine differing trends in real cost growth over time. The most important observation from this analysis is the variation in labour cost movements and materials cost movements since 1997. The Chart clearly demonstrates that, particularly since 1997, labour and plant costs have increased sharply in real terms whilst construction materials costs have, on average, fallen overall, and in two of the three periods.

Labour costs (rather than Plant) are the main component of the Labour & Plant index, rising above trend in response to increasing industry workload and the long term restructuring of wage agreements involving changes to bonusing arrangements. At the same time, materials cost increases have been generally restrained as input costs for manufacturing have generally remained subdued.

Chart 2.3 Labour & Plant and Materials Components of HOCOS

A similar pattern of Labour & Plant and Materials cost movements was also observed in respect of the Resource Cost Index of Building: Non-Housing (NOCOS).

The study showed that there was a slightly greater fall in the real cost of materials for non-housing work (-5.4%) than for housing work (-3.5%) since 1998. However this small divergence in costs is not apparent in Producer Price Indices data compiled by ONS in respect of Construction Materials and House-building Materials. This data shows that the movement in prices for construction and house-building materials was exactly equal between July 1998 and January 2003.

2.3 Building Cost Information Service

2.3.1 General

The Building Cost Information Service (BCIS) publishes a series of Input Cost indices for the UK, which serve the same function as the Resource Cost Indices published by the DTI, tracking changes to contractors' input costs (labour, materials and plant). The indices are based on cost models of 'average buildings' compiled by BCIS, for new build schemes. The inputs to the indices are the same as those used by the DTI, Work Category indices for use with Price Adjustment Formulae for Construction Contracts. The weightings of the various indices within the cost models are published by BCIS. The BCIS series of indices are shown in Appendix A.

In addition to the General Building Cost Index, BCIS publish a series of indices representing alternative construction methods and indices for component parts. The indices representing different construction techniques are derived from different cost models:

• Steel framed, concrete framed and brick built construction.
• General building work, mechanical, electrical and lift services.
• Labour, plant and materials.

The BCIS General Building Cost Index can be directly compared to the DTI index series. In the context of the study, although there is no discrete housing index, the isolation of masonry construction in a specific separate index makes that series likely to be the most appropriate for use in connection with the social housing sector, as the inputs to the 'brick built construction' model have been based largely (but not entirely) on housing projects.

As the DTI and BCIS Building Cost indices are derived from broadly similar data sources, the analysis checked to see if these series are comparable.

2.3.2 Observations

Comparisons were made between DTI and BCIS cost index series to determine whether the index series were consistent or divergent. It was found that the two sets of indices had a strong correlation, reflecting the use of common data inputs.

As a result it is reasonable to assert that relationships can be inferred from one to the other. This is valuable, as the DTI's HOCOS index and specific building type indices produced by BCIS, can be used to determine an appropriate input cost trend for the Scottish Social Housing market.

2.3.3 Cost indices for building types

The BCIS General Building Cost Index is a compilation of different building types. As well as the BCIS General Building Cost Index, BCIS produce separate index series based on models for:

• Steel framed buildings
• Concrete framed buildings
• Brick buildings

Over the whole period, the costs of brick buildings have risen a little ahead of the framed construction types, particularly over the last five years. This pattern reflects restraint in prices of both structural steelwork and steel reinforcement (for concrete frames). Neither structural steelwork nor steel reinforcement play a major part in Scottish Social Housing repair, improvement and maintenance work nor for the majority of new build work. Therefore the brickwork construction index should be considered as the more appropriate of the three sub-indices of the BCIS cost index family.

2.3.4 Cost indices for building and services

BCIS UK index series were also separately examined for:

• Building work
• Mechanical and Electrical work

The pattern of movement was similar to that observed from the DTI family of indices. The cost of building work fell, in real terms, between 1990 and 1992 while mechanical and electrical costs did not. As building costs recovered between 1992 and 1996, mechanical and electrical costs increased by a similar amount. Between 1985 and 1996, building costs show no overall change in real terms, while M&E costs rose by 8%. Since 1996 the pattern has been reversed, M&E costs displaying no variation from general inflationary trends while building costs have risen by 5% in real terms.

The differential trends of M&E and building work costs have largely been due to labour wage awards. In the mid 1990's greater union strength on the part of electrical and heating and ventilating engineers secured larger annual wage increases for their members than on the building side (see Chapter 9). However, since 1999, while the Services unions have continued to secure above inflation pay awards, building operatives have negotiated more substantial wage awards as part of a process of 'catching-up', with union rates rising to more closely reflect site pay.

2.3.5 Cost indices for labour, plant and materials

In common with the DTI, BCIS produce further sub-indices for labour, materials and plant costs. The pattern, as expected, is similar to that identified in the DTI data and the commentary related to Chart 2.3 above applies here. BCIS have a separate index for plant costs, the movement of which bisects the labour and materials costs trends. However, plant is not a major cost item in social housing repairs, improvement and maintenance work and this is not a significant cost driver.

2.3.6 'Building' Housing Cost Index

BCIS do not compile a separate housing cost model as a sub-set of their General Building Cost Index. However BCIS have compiled a Housing Cost Index since the 1970's, originally published in 'Building' magazine, and still known as the 'Building' Housing Cost Index. This uses national wage agreement data for its labour component but privately collected information on materials costs, rather than relying on ONS statistics.

Over the period from 1985, the 'Building' Housing Cost Index has risen in real terms by more than both the DTI's HOCOS index and the BCIS Brick Construction Cost Index, increasing 13% overall compared to 3% and 7% respectively. A detailed examination of the data suggests that, particularly from 1999 onwards, the labour element of the index, apparent from an annual step change on examination of the quarterly figures, appears to have driven the index up at a higher rate.

The index had a base labour weighting (builders, plumbers and electricians) of 56% (excluding site and head office overheads) in February 1978. The BCIS Brickwork Construction Cost Index had a much lower base labour weighting of 38% in June 1976. The conclusion must be that the DTI HOCOS index has an even lower labour weighting. However, as DLE have been unable to secure details of the weightings of the DTI indices, it has been unable to confirm that this is the case.

Refurbishment work is likely to be more labour intensive than new build work and, consequently, the 'Building' Housing Cost Index may be a valid index to monitor. However the index is no longer published in 'Building' magazine or anywhere else though it is available from BCIS on subscription or on-line (to On-line subscribers).

2.4 Conclusion

2.4.1 Summary of results from published index series

Table 2.1 summarizes the results of the various studies undertaken on Capital Cost data. The comparison of resource cost data from BCIS and DTI sources over the period 1985 to 2002 shows that both measures generated broadly similar results.

Table 2.1 Summary of Capital Cost Indices and Real Percentage Changes from Published Index Series

The principal findings derived from this analysis are:-

• Real construction input costs grew by between 0.2 and 0.4% per annum overall from 1985 to 2002, then subsequently fell by between -0.4 and -1.0% per annum between 1985 and 1992 before growing again by between 0.8 to 1.0% per annum over the last ten years.
• Real labour costs have risen at a faster rate than real materials costs in all periods except between 1992 and 1997.
• Labour costs have risen in real terms throughout the study period and have risen by 28% overall.
• Materials costs have fallen in real terms throughout most of the period, except between 1992 and 1997. Over the whole period, materials costs have fallen by 10 to 14%.
• Over the whole period, there has been no substantial difference in cost movement associated with building type, all the BCIS and DTI options rising between 0.2 and 0.4% per annum (with the exception of the 'Building' Housing Cost Index).
• Between 1985 and 1992 all the variants experienced falling real costs (-0.4 to -1.0% per annum); between 1992 and 1997 all increased consistently at 0.8 to 0.9% per annum; over the latest period, 1997 to 2002, average annual increases varied from 0.5% for BCIS steel framed buildings to 1.3% for BCIS brick buildings, though the 'Building' Housing Cost, with its higher labour weighting, rose at an average rate of 2.3% per annum.
• From 1997 masonry built buildings (DTI HOCOS and BCIS Brick Construction Cost Index) rose at the higher rate of 1.2 to 1.3% per annum while the other building types (DTI NOCOS and BCIS Framed Construction Cost Indexes) increased at the lower rate of 0.5 to 0.8% per annum. See also note in paragraph above regarding the 'Building' Housing Cost Index.
• Labour costs have been the major driver behind variations in real building cost trends, shown in the breakdown of the index series into component parts.

2.4.2 Trends in U.K. materials costs

Differing patterns of real cost change between labour and materials costs are clearly discernible from examination of the sub-series indices. Real UK cost changes for materials have been consistently falling over most of the 17-year study period, for both housing and non-housing work, declining by up to 14% since 1985 (based on DTI figures). The main reasons behind this trend are likely to be increased investment by materials manufacturers and suppliers, improved labour productivity, greater competition from imported products and, possibly, subject to further research, long term declining profitability of suppliers.

2.4.3 Trends in U.K. labour costs

By contrast, labour costs, though declining relative to general inflation during the first part of the study period (according to DTI indices), increased sharply, particularly since 1997, with the UK figure rising 21 to 28% (in real terms) since that year. The overall increase in housing costs compared to non-housing work is due to a higher labour weighting involved in housing projects than in construction in general. It should be noted that pressure on labour costs in labour intensive elements such as masonry construction and wet finishes such as plastering is likely to continue to increase in real terms for the foreseeable future (while construction activity remains at or above current levels).

However it is likely that productivity gains and greater use of prefabrication since the base date of 1985 have resulted in less labour being required on site to produce a given level of output than is implied by the weights used to construct the index. This means that the series may have overestimated the 'true' rate of 'building cost per unit' change in recent years. This should be borne in mind when considering projections in the costing of schemes.

2.4.4 U.K. long term building cost trends

In the costing of long term projects, the evidence suggests that (at UK level) capital costs can reasonably be assumed to increase in line with general inflation. In the costing of shorter term projects, positive (or negative) real cost changes should only be assumed if there is strong demonstrable evidence to suggest that the construction industry is in or is about to enter a period of relative expansion (or contraction) or if there is incontrovertible evidence that local factors will similarly influence real cost trends.

2.5 Scottish Social Housing Refurbishment Cost Model

All of the index series considered so far have been generic, relating to UK national averages for new build work. Although the cost patterns identified have been largely uniform, for the purposes of this study, it has been important to develop a cost model specific to the nature of the work under consideration. In response to this need, DLE have created a Social Housing Refurbishment Cost Model specifically for this study.

Using capital expenditure profiles for component replacement for housing, taken from the plans of two Scottish Councils, one predominantly rural and one urban, DLE have created a Social Housing Refurbishment Cost Model, which can be tested against the more generic input cost indices produced by DTI and BCIS.

Separate models were developed for each council's housing stock, providing weightings based on the 60 Building Work categories of the DTI's Price Adjustment Formulae for Construction Contracts or to the additional work categories associated with heating and ventilating and electrical work.

The index series and resultant trend lines have been calculated back to 1991, the first full year of the latest series of Price Adjustment Formulae Indices

The trend lines for the expenditure of the two separate councils are for the most part almost inseparable, even though they represent refurbishment, in one case of rural housing and, in the other, urban housing stock. As the trend lines of the expenditure of the Councils' stock are so similar, subsequent comparison work was undertaken using an average index derived from the two models.

Clearly, there is no guarantee that the housing stock of the two councils sampled is typical of Scottish social housing and DLE have not attempted to compare the mix and condition of this housing with the remainder of the country's stock. However the close correlation between the two index series provides confidence that the DLE refurbishment model should be representative of a much wider sample.

Chart 2.4 plots the cost trend of the Scottish Social Housing Refurbishment Cost Model Index with the DTI's Resource Cost Index of House Building (HOCOS), the published index with the closest match to the mix of work under consideration.

Chart 2.4 Comparison of Scottish Social Housing Refurbishment Cost Model Index and HOCOS

Over the study period (1991 to 2002), the two index series have shown a similar overall real cost increase, HOCOS rising 8.6% (average 0.8% p.a.) and the DLE cost model 11.9% (average 1.0% p.a.). However between 1992 and 1997 the Scottish Housing Cost Model increases at a slower rate 0.4% p.a., compared to 0.8% p.a. for HOCOS, while between 1997 and 2002 the reverse is true, with the Cost Model rising by 2.2% p.a. compared to 1.2% p.a. for HOCOS.

It is not possible to firmly establish the reasons for this variation, as DTI have not been able to provide details of the weightings used for HOCOS. However, based on conclusions from earlier analyses it is likely that the higher increase in the Social Housing index after 1997 is related to a greater weighting for items with a higher labour content, and the higher proportion of labour in the index per se. The higher real cost increase follows wage trends in the construction industry exceeding the general level of earnings increases over this period as a result of rising construction workload exerting pressure on a finite labour supply.

It should be noted that, over the later period considered, 1997-2002, the Cost Model has increased in real terms by a very similar percentage to the 'Building' Housing Cost Index.

The conclusion from this analysis is that:-

• Building costs in the new build and refurbishment sectors of social housing are, in the long run, broadly comparable.
• Refurbishment and component replacement costs have risen over the short-term at a faster rate (2.2% per annum in real terms, based on 1997-2002), in response to higher wage increases.
• It is appropriate to consider refurbishment and component replacement workload as having distinct characteristics that cannot be fully reflected in the conventional HOCOS and BCIS published indices.
• The 'Building' Housing Cost Index may most closely reflect the movement of the Refurbishment Cost Model Index but, given the relatively similar movement overall of the specifically generated cost model and HOCOS over the period examined, HOCOS can also be recommended as a reasonable published substitute index.

« Previous | Contents | Next »