Final Report Deriving Measures of Plant-level Capital Stock in UK Manufacturing, 1973-2001a Submitted to the DTI by Professor Richard Harris (Independent Economic Consultancy Services Limited) a This work contains statistical data from ONS which is Crown copyright and reproduced with the permission of the controller of HMSO and Queen's Printer for Scotland. The use of the ONS statistical data in this work does not imply the endorsement of the ONS in relation to the interpretation or analysis of the statistical data. i [this page is blank] ii Executive Summary E1. In order to estimate total factor productivity (TFP) (the preferred measure of productivity vis-à-vis labour productivity) and other supply-side analysis based on the use of production functions, reliable measures of capital stock a are a necessary prerequisite. E2. Currently, the capital stock series used by the Business Data Linking (BDL) section at the ONS measures capital stock using what is known as Reporting Unit (RU) data, and various methods based on the perpetual inventory approach. The series were computed by Martin (2003). The use of RU data, and the assumptions used to derive capital stock series need to be carefully analysed, and the drawbacks and limitations considered. This is the major purpose of this study, to consider such issues. E3. An accurate measure of the capital stock that is intended for use in estimating production relations should represent the total amount of capital services available for producing output. This means taking into account efficiency losses due to deterioration (including obsolescence). This requires the use of an appropriate rate of deterioration of capital goods that reduces their efficiency through time as they are used to provide capital services and as they age. E4. It was argued by Denison (1972) that the expected time pattern of deterioration of a capital good is expected to exhibit a slow rate of deterioration at the beginning, becoming more rapid as the expected length of life of the good approaches. Thus gross and net (of straight-line deterioration) plant and machinery capital stock figures are calculated in this study using the perpetual inventory method and the expected length of life of capital goods adopted by the ONS since 1983. Once gross and net stock figures have been calculated, the i approach used by Denison (1972) of weighting these in the ratio of three to one is adopted to obtain net stock figures which incorporate the desired pattern of deterioration, at first slow, followed by more rapid deterioration. E5. Other authors have advocated the geometric depreciation pattern as an alternative approach to the Denison approach method used in this study. However, this implies a far higher rate of deterioration than is likely in economic terms, involving a much higher rate of capital-embodied technical progress than those typically found in the empirical literature. E6. Thus, the Denison approach with ONS asset lives was used, with the perpetual inventory approach, to produce plant and machinery capital stock estimates using gross investment data for plant and machinery taken from the Annual Respondents Database (the ARD). E7. The technical details of how this was done are presented, along with the statistical routines required to estimate UK manufacturing plant and machinery capitals stocks for each plant in operation between 1970 and 2001. E8. The list of issues that were then considered include the following: 1. Whether plant (LU) or establishment (RU) level data should be used 2. Implementation of the Perpetual Inventory method and specifically: i. What depreciation series should be used? ii. What start values are needed (i.e. the first year of gross investment data used) in order to derive an accurate series iii. The availability of plant level gross investment data (acquisitions net of disposals) iv. How to take account of plant closures E9. These issues were considered mainly by contrasting the preferred approach used in this report with the approach used by Martin (2003). Various evidence is presented to show that the methods used here to calculate the plant and ii machinery capital stock have advantages over those employed by Martin since they: • are based on LU data and thus avoid the problem that LU’s can be reallocated between RU’s for accounting purposes by firms, or when plants are bought and sold, or when they are opened and closed; • the Denison approach to calculating economic deterioration seems more realistic when compared to using a geometric depreciation series, especially if the purpose of calculating capital stock estimates is to use these when estimating a production function or when estimating total factor productivity; • post-1969 data at the plant level is used here, while Martin uses post-1979 data, and thus the influence of benchmark data is lower in the present estimates, especially from 1980 onwards. The benchmark series used here were also more disaggregated, and thus likely to have been more accurately distributed to those plants that were open in 1970 (and thus required pre-1970 capital stocks to be allocated to them), especially as Martin uses the average value of intermediate inputs of RU’s based on their entire operating life to allocate the benchmark data (rather than investment and employment shares linked to the benchmark period); • investment data is interpolated (when necessary) using a population estimate of investment in a 5-digit industry in any particular year (after subtracting investment already identified to belong to those plants that were selected for inclusion in the ABI), while Martin simply interpolates linearly between observed investment levels for each RU; • the capital stock produced here takes account of plant closures, while Martin does not. iii E10. However, ‘spreading back’ investment data from selected RU’s based on LU employment shares does introduce an unknown level of bias into the capital stock estimates produced at the plant level, because of inaccuracies in plant level employment estimates for the smallest firms. Improvements in the accuracy of measuring the capital stock using the perpetual inventory approach would be achieved if the ONS were to consider the following changes: • collecting information on capital expenditure at the local unit level (in addition to employment data) when conducting the ABI; • increasing the frequency with which LU employment estimates are updated for smaller firms. The National Assembly for Wales has for some years funded a top-up to the ARI for the region and Scotland has recently agreed the same with the ONS. • obtaining more up-to-date estimates of economic depreciation of assets by industry, in order to provide better estimates of the rate of deterioration and obsolescence. iv Table of Contents Executive Summary i Chapter 1: Introduction 1 Chapter 2: Measuring the Capital Stock 7 Chapter 3: Plant Level Estimates of the Capital Stock 17 Chapter 4: Issues in Measuring the Capital Stock 25 References 39 Statistical Appendix 42 v [this page is blank] 1 1. Introduction 1.1 Given the availability of the panel micro-level database on enterprises in UK manufacturing comprising the Annual Respondents Database (the ARD), various analyses relating broadly to issues such as productivity and growth can now be undertaken. This is directly relevant to the building up of the evidence base needed by government in this area in order to achieve its goal of boosting UK productivity levels and overall GDP. 1.2 With such micro panel data it is possible to measure such variables as labour productivity, and efficiency, and consider related issues such as entry and exit, and the impacts of changes in ownership. All of these topics have been previously looked at (sometimes for all UK manufacturing but also separately for different sectors) by Harris and others.1 1.3 However, in order to undertake studies related to total factor productivity (TFP) (a preferred measure of productivity vis-à-vis labour productivity – see below) and other supply-side analysis based on the use of production functions, reliable measures of capital stock are a necessary prerequisite. How the capital stock is measured in terms of the use of the perpetual inventory method, and the data needed to undertake such calculations, results in a number of significant and important issues that have to be considered. 1.4 Currently, the capital stock series used by the Business Data Linking (BDL) section at the ONS measures capital stock using what is known as Reporting 1 For a detailed description of the ARD see Oulton (1997), Griffith (1999), and Harris (2002). Analysis using the database covers a range of areas; cf. Disney, Haskel and Heden (2003a,b), Harris and Drinkwater (2000), Harris (2001), Harris and Collins (2002), Harris and Robinson (2002, 2003, 2004a,b), and Harris and Hassaszadeh (2002). The counterpart to the ARD in the US is the Longitudinal Research Database – or LRD - for US manufacturing provided through the US Bureau of Census. This has been analysed fairly extensively in recent years, covering various areas linked to productivity (e.g., Bahk and Gort, 1993; and Bartelsman and Dhrymes, 1998); the impact of ownership change on productivity (McGuckin and Nguyen, 2001); capital efficiency (e.g., Doms, 1996) and entry and exit (e.g., Doms, et. al, 1995; Olley and Pakes, 1996; Kovenock and Phillips, 1997) 1 Unit (RU) data, and various methods based on the perpetual inventory approach (see Martin, 2003).2 The use of RU data, and the assumptions used to derive capital stock series need to be carefully analysed, and the drawbacks and limitations considered. Evidence is presented here (based on Harris and Drinkwater, 20003) showing that Local Unit (LU) data should be preferred, as well as different assumptions regarding lengths-of-life of assets, how they depreciate, how data is weighted, and historical information on gross investment at the plant (i.e. LU) level. 1.5 Therefore this project looks at the different methods used to calculate the capital stock (concentrating on the BDL approach and that used by Harris and Drinkwater, op. cit.), and the consequences of using these approaches. In addition, an up-to-date series of plant level (plant and machinery) capital stock has been compiled for the ONS for inclusion in the BDL database, along with the statistical algorithms that can be used to update the series when new ABI data becomes available to the ARD. Those proposing to calculate their own capital stock measures or to update the data produced from this report are advised to consult existing documentation on the ARD available in the BDL at the ONS. 2 Others have calculated their own estimates of capital stock using similar approaches to Martin (e.g. based on RU data, a benchmark series often starting in 1980 or later, and an assumption that depreciation has an exponential distribution). Examples include Griffith (1999) and Disney et. al. (2003a, 2003b). Given that their approaches are very similar and/or less detailed than Martin’s, only the latter is considered in any detail here. 3 This current study uses the same approach as that employed by Harris and Drinkwater, but updates the series to cover 1994-2001. It also differs from the previous study in that it compares the approach used to that employed by Martin (and others – see footnote 2). 2 Measuring productivity – the preferred approach 1.6 As stated in par. 1.3, capital stock measures are necessary in order to obtain estimates of total factor productivity rather than, say, labour productivity. This sub-section sets out why measures of TFP are preferred vis-à-vis labour productivity. It is useful to start with a simple Cobb-Douglas4 production function: y = α 0 +α E e +α M m +α K k +αT t (1.1) where y refers to the logarithm of real gross output; e refers to the logarithm of employment (with αE measuring the elasticity of output with respect to employment – i.e.∂y/∂e); m refers to the logarithm of real intermediate inputs (with corresponding elasticity of output, αM); and k refers to the logarithm of capital stock (with corresponding elasticity of output, αK). Totally differentiating this function with respect to time to obtain rates of change (and expressing terms such as dy / dt as y& ), a measure of total factor productivity growth can be obtained as: TF&P = α T = y& − α E e& − α M m& − α K k& (1.2) Note, this measures the increase in output that is not attributable to increases in e, m, or k; rather it measures the contribution to growth of all influences other than the factors of production, capturing such determinants as technological progress and/or increases in efficiency (note, inefficiency also captures any under-utilising of factor inputs). Thus, if any factor input increases, then via 4 Other forms of the production function (e.g. CES or translog), or indeed a general function, could be used, and the points to be illustrated will not alter in any significant way. 3 (1.1) output increases by a value depending on the elasticity of output with respect to the factor increasing, and the impact on TFP in (1.2) is zero. 1.7 In terms of labour productivity growth, a relationship can be obtained by subtracting the logarithm of employment from both sides of (1.1) and expressing the result in terms of rates of change with respect to time: y& − e& = (α E − 1)e& + α M m& + α K k& + TF&P (1.3) This shows that increases in labour productivity ( y& − e&) are negatively related to increases in employment [since (αE−1)<0], and positively related to increases in intermediate inputs, capital stock and TFP. Indeed, if over time there is an increase in capital deepening (cet. par. the K/E ratio rises as capital is substituted for labour perhaps due to greater automation) or outsourcing (cet. par the M/E ratio rises as less is made internally and more semi-finished and finished products, and services, are bought from suppliers), then labour productivity will increase as relatively less labour is used to produce output.5 Thus, increases in labour productivity do not depend on just technological progress and/or gains in efficiency, since what happens with the other factors of production is also important. 1.8 Figure 1.1 provides some insights with regard to what has been happening in UK manufacturing in recent decades;6 labour productivity (based on the gross output measure) rose fairly constantly throughout the 1973-1998 period; however, this was likely to have been at least in part due to capital deepening 5 If a value-added production function were used instead of a gross output function (with VA=Y−M), and constant returns-to-scale imposed with perfect competition in factor and output markets, then (1.3) simplifies to: y& − e& = (1 − α E )(k& − e&) + TF&P (1.3a) which shows that labour productivity growth depends positively on capital deepening and TFP growth. 6 Note the data in Figure 1.1 are based on (population weighted) totals for UK manufacturing, not averages of plant or firm level data. 4 (increases in K/E), particularly in the 1970’s and early 1980’s, and outsourcing (increases in M/E), which seems to have increased at a more significant pace from the mid-1980’s onwards. Figure 1.1: Real outputs and inputs in UK manufacturing, 1973-1998 3.30 2.80 gross output per employee capital per employee intermediate inputs per employee M/E 1973=1 2.30 Y/E 1.80 K/E 1.30 0.80 1973 1974 1975 1976 1977 1978 1979 1980 1981 1982 1983 1984 1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 Source: based on own calculations using the ARD 1.9 In conclusion, the major weakness of using a measure of labour productivity growth that is primarily intended to capture the impact of improvements in technology and/or efficiency is that it is also significantly influenced by substitution between the factors of production. In contrast, TFP measures capture the ‘pure’ impact of shifts in the production possibility curve (due to technical change) or movements towards it (due to improvements in efficiency). 1.10 It can also be seen that there is an issue of which labour productivity measure should be used (i.e., based on gross output or gross value-added).7 The answer depends on whether the production function is separable or not (i.e., whether intermediate inputs be separated from K and E in the formulation of the 7 Of course, this same question arises when calculating TFP – should a gross output or value added production function be used? 5 function) – and the empirical evidence strongly suggests not – while the literature on outsourcing suggests that the gross output approach captures more exactly measures of efficiency while value-added functions are more closely related to profitability (Gorzig and Stephen, 2002). 6 2. Measuring the Capital Stock 2.1 An accurate measure of the capital stock that is intended for use in estimating production relations should represent the total amount of capital services available for producing output. This means taking into account efficiency losses due to deterioration (including obsolescence). However, there is a second 'economic accounts' measure of the capital stock that takes into account capital that is 'used up' in production; that is, depreciation. Triplett (1996) provides a comprehensive account of the differences between these two concepts of the wealth and productive capital stock, which has been the source of much debate in the literature. Essentially, Triplett (op. cit., p. 100) shows that while deterioration and capital used up in production (depreciation) have similarities, they are not equal. The main difference between them is that deterioration measures the current services lost from using the capital stock in any one period, and thus the loss in potential for next period's production; depreciation measures (in value terms) the lifetime deterioration in the capital stock that reduces the long run income flow from that stock. That is, deterioration refers: "… to the loss of capital services in the following period that arise from the use of the capital good in any period, t + i" while depreciation is "… the diminution of the total stock of services embodied in a capital good because it has been used in production for one period" (p. 100, italics from original). Triplett notes that: Capital used up is not the appropriate concept for production analysis… deterioration is not the appropriate definition for the purposes of economic accounts… These are not mere semantic 7 differences, or issues that arise from differences in the use of language; they are conceptual distinctions that arise because the production function use of capital and the economic accounts use are different purposes (p. 103) 2.2 Here we concentrate on a 'production analysis' concept of the capital stock that takes account of deterioration. This requires the use of an appropriate rate of deterioration of capital goods that reduces their efficiency through time as they are used to provide capital services and as they age. Essentially, we define the net stock of capital as8: t K(t) = E (υ ) ∫ E (t ) . D(t − υ ). I (υ ). dυ (2.1) −∞ where I(ν) is constant price gross investment of year ν, weighted by both physical decay D between time ν and t, and by accumulated embodied technical progress (obsolescence) E between time ν and t. Therefore, vintages installed in year ν provide less capital services as the age of I(ν), i.e., t − ν, becomes sufficiently large due to (i) wear and tear, resulting in D → 0,9 and (ii) accumulated technical progress means that the newest vintages in time t embody more capital services than do older fixed assets, thus over time E(ν)/E(t) → 0. Physical decay and obsolescence combine to produce deterioration. Therefore, information is required on the rate of deterioration that reduces the capital services available from any vintage and eventually removes the ageing asset from the remaining capital stock. 8 A simpler way of thinking of the capital stock is based on the perpetual inventory formula: K t = (1 − δ ) K t −1 + I t where K is the capital stock, δ is economic depreciation, and I is investment. D(t − ν) can be thought of as an index of decay equal to 1 when t = ν, but becoming equal to 0 when t − ν becomes sufficiently large. 9 8 2.3 Ideally, we require empirical estimates of the level of known deterioration to enable us to calculate both the rate (or pattern) of deterioration and the value of the actual length of life of a capital good. Unfortunately, this information is not available and therefore some other technique must be used to estimate the expected life of a capital good, which provides an indication of the period over which deterioration must take place. The perpetual inventory method is typically adopted for this purpose, with the rate of deterioration assumed to follow some simple pattern such as straight-line or exponential decline. If rates of physical decay and embodied technical progress are assumed to be stable over long periods of time, then the expected length of life will remain the same and using expected length of life is an accurate indication of how long deteriorating capital goods remain in a net stock measure. However, these assumptions appear more appropriate for physical decay, and not technical progress, and this is presumably the major reason why the ONS adopt the procedure of shortening asset lives every five years for plant and machinery expenditure undertaken since 1950.10 The length of life assumptions do not take account of premature scrapping, which is mainly caused by obsolescence. This occurs when a capital good ceases to earn a positive quasi-rent, often caused by a lack of demand for the firm’s product or a sharp change in relative factor prices, and is scrapped in advance of the period when repairs to combat physical decay become too great. 2.4 As to the discussion over the appropriate rate of deterioration, it was argued by Denison (1972) that the expected time pattern of deterioration of a capital good 10 Asset lives decline depending on which 5-year period the asset was purchased between 1950 and 1970 (assets bought since 1970 are assumed to have the same lengths of lives as 1970 assets). For example, the average life assumption for all manufacturing plant and machinery declined from around 29 years in 1949 to 22 years by 1970. 9 is expected to exhibit a slow rate of deterioration at the beginning, becoming more rapid as the expected length of life of the good approaches.11 The reasons for this are that firms typically undertake maintenance and repair in order to maintain the same performance level as when the machine was new (Jefferson, 1971), with this activity increasing with the age of the capital goods, while the effect of obsolescence on the rate of deterioration is probably small (Barna, 1962).12 This assumption is more applicable when the rate of capital-embodied technical progress is low over time since equation (1) shows that obsolescence not only affects the length of life of an asset but it also makes older vintages less productive than newer ones. Thus lowering the length of life of assets, on the assumption that technical change has increased in more recent years is presumed to reduce any impact of obsolescence on the rate of deterioration. 2.5 Given the above discussion, gross and net (of straight-line deterioration) plant and machinery capital stock figures are calculated using the perpetual inventory method and the expected length of life of capital goods adopted by the ONS since 1983.13 As set out in Hulten and Wykoff (1996 p.15), if ϕs represents the efficiency (or remaining capital services) of an s-year-old asset as a function of the productive capacity of a newly produced asset, then the deterioration pattern for the gross stock measure is: ϕ0 = ϕ1 = …. = ϕT-1 = 1, ϕT+t = 0 t = 0, 1 ,2 …… (2.2) 11 Note, Denison referred to depreciation rather than deterioration, but failed to distinguish between the production analysis use of capital stock and the economic accounts measure. 12 Barna (1962) is concerned with capital stocks in the 1950s, when there was a large amount of heavy industry. 13 Note, there are up to 6 different classes of assets covering plant and machinery capital goods purchased by each of the 18 industries, each with a different average length of life (which also changes over time although post-1970 lives are used throughout in this study). Table A2.1 in the appendix presents the lengths of life assumptions used since 1970. Gross investment figures by industry were available for the 1948 to 1969 period and individual plant level estimates were available from 1970; see the next chapter for details, including how the two datasets were merged. 10 That is, assets retain full efficiency until retirement at time T, the service life of the asset. If straight-line deterioration is adopted then the pattern for the net stock is: ϕ0=1, ϕ1=1-(1/T), ϕ2=1-(2/T), … ϕT-1= 1-[(T-1)/T], ϕT+τ = 0 τ = 0, 1 ,2 2.6 (2.3) For completeness, choosing the exponential distribution, deterioration (and in this instance depreciation) would occur at a constant rate δ=(ϕt-1-ϕt)/ϕt-1 so that: ϕ0=1, ϕ1=(1-δ), ϕ2=(1-δ)2, …. ϕt=(1-δ)t (2.4) Note, the exponential distribution is not (directly) determined by the service life of the asset, and only asymptotically declines to zero. Thus under the exponential distribution the efficiency function and the age-price function are identical and deterioration and depreciation are equal. 2.7 Once gross and net stock figures have been calculated, the approach used by Denison (1972) of weighting these in the ratio of three to one is adopted to obtain net stock figures K(t), which incorporate the desired pattern of deterioration, at first slow, followed by more rapid deterioration.14 2.8 Other authors have advocated different deterioration and/or depreciation patterns or an alternative approach to the perpetual inventory method used in this study.15 For instance, Oulton and O'Mahony (1994) use an exponential rate of depreciation (which is equivalent to exponential deterioration) together with 14 It is possible to use weighting ratios other than the preferred 3:1 ratio. However, other ratios would move the distribution closer to the distribution for the gross stock or closer to straight-line deterioration, while it is argued here that the Denison approach is more appropriate. Figure 2.1 illustrates this point. 15 Those interested in an economic accounts measure of capital stock need to measure depreciation. As pointed out in the text, only when the exponential distribution is used do depreciation and deterioration coincide. Thus using the exponential distribution results in an internally consistent measure of capital stock in that the two concepts are then the same (see Jorgenson, 1996, who gives this argument in favour of using the exponential distribution). 11 the ONS length of life assumptions.16 They justify the use of the exponential distribution with reference to Hulten and Wykoff (1981), in which the prices of second-hand assets were found to decline geometrically with an asset’s age in the US.17 Oulton and O'Mahony (1994) nevertheless argue that the rise in efficiency of new assets (or equivalently the increased obsolescence of older ones) leads to an overall geometrically declining deterioration pattern even though they accept that for plant and machinery, physical deterioration is unlikely to follow such a pattern (the ‘light bulb’ or one-hoss shay pattern being more likely). Advocates of the exponential distribution usually favour its use for reasons other than just to take account of obsolescence; e.g., because its is consistent with the economic accounts definition of the capital stock; it is implied by second-hand US capital price data; and it also incorporates various factors that can lead to a wide-band distribution of retirements in practice (e.g., the loss of assets due to fires, explosions, pre-mature scrapping). Hulten and Wykoff (1996) state: …it may well be true that every single asset in a group of 1000 assets depreciates as a one-hoss shay, but that the group as a whole experiences near-geometric depreciation. This fallacy of composition arises from the fact that different assets in the group are retired at different dates; some may last only a year or two, others ten to fifteen years. When the experience of short-lived assets is 16 The approach taken by Martin (2003) when calculating the BDL capital stock series was also to use the exponential rate, and an average rate of 0.06 for plant and machinery (see Chapter 4 for a discussion and implications). 17 Such data is not available for the UK, and we would argue that it would not necessarily reflect accurately both wear and tear and obsolescence. This is because second-hand asset price data reflects the impact of depreciation and not just deterioration (the former taking account of deterioration over the entire life of an asset), and it has been argued that using used-asset market price data as an indicator of in-use asset values is problematic if the relatively small number of assets resold in second-hand markets are not of as ‘good quality’ as those assets that remain with the plants that undertook the initial investment. 12 averaged against the experience of long-lived assets, and the average cohort experience is graphed, it will look nearly geometric (p. 18). 2.9 However, this implies that a number of assets are pre-maturely scrapped, destroyed or ‘lost’. That is, many assets have to be short-lived in order to produce an exponential deterioration rate, especially if (as in this study) premature scrapping due to closure is accounted for separately. 2.10 A simple example demonstrates the implications of using the exponential distribution. If an asset has an average service life of 20 years, then after five years it will typically offer only 33 per cent of the capital services that would be available from a new asset. After ten years, only 11 per cent of the asset’s initial services are available, falling to 3 per cent in year 15. In other words, when deterioration is assumed to be very high in the first few years after installation, a new asset is three times more productive than an asset that is one-quarter of the way through its life. This implies a far higher rate of capital-embodied technical progress than those typically found in the empirical literature (e.g., Kalt, 1978, estimated capital-embodied technical progress at 0.01 per cent per annum for the USA over the 1929-67 period, while Hulten, 1992, reports a figure of 0.3 per cent per annum for 1949-83). 2.11 An alternative approach to the perpetual inventory method that has been suggested is one which centres on the use of a putty-clay model of production (e.g., Coen, 1980; and Anderson and Rigby, 1989). Basically this means using an investment model to impute the ‘best’ rate of depreciation and service life that fits gross investment data matched to employment levels through time. Besides the untested hypothesis that once installed a capital good requires a 13 fixed amount of labour to operate it, the approach is also subject to other theoretical and empirical limitations (see Fabricant, 1980 for a discussion).18 Figure 2.1: Different distributions for deterioration: Iron and Steel industry 100 Net Stock Gross Stock Denison Exponential 90 80 initial investment in year 0 = 100 70 average service life = 24.6 years 60 50 40 30 20 10 46 44 42 40 38 36 34 32 30 28 26 24 22 20 18 16 14 12 8 10 6 4 2 0 0 Years from original investment 2.12 To illustrate the use of the Denison deterioration pattern instead of the more frequently used exponential pattern, Figure 2.1 shows various outcomes for the Iron and Steel industry following the purchase of a single capital good (costing £100 in real terms). The ONS assumes that investment in any year is subdivided (in this industry) into 4 sub-groups, each accounting for a different proportion of the total investment and each with a different service life (see Table A2.1 for details).19 For example, 77.9 per cent of the investment is 18 Note, the assumption of ‘putty-clay’ technology is not essential for this type of approach. For instance, Nadiri and Prucha (1996) jointly estimate a labour-demand equation, based on a normalised variable cost function in which Kt is a determinant, with the following identity: Kt ≡ It + (1 − δ)Kt-1, where It denotes gross investment and δ the depreciation rate of capital. The authors found that δ took a value of 0.059 using data on US manufacturing plant and machinery gross investment for 1960-88 (and an initial benchmark estimate of the capital stock). 19 In theory, it is possible to use different lengths of life assumptions (and thus different rates of deterioration). However, we only have information collected by the ONS on which to base asset lives, with no evidence that alternative estimates should be used in their place. 14 presumed to belong to an asset class where the length of life is 26 years. For Iron and Steel, the average length of life when combining all four sub-groups is 24.6 years. For the net and gross (and thus Denison) stock, assets are assumed to be retired 10 per cent either side of the average length of life of each sub-group. The exponential pattern is based on the ONS length of life of each sub-group and the declining balance rate (DBR) derived from Hulten and Wykoff (1981) with δ = DBR ÷ T.20 2.13 Figure 2.1 shows that when the asset is 6 years old (approximately one-quarter of the way through its average length of life), none of the gross stock has deteriorated, 89.9 per cent of the Denison stock remains, 69.8 per cent of the net stock is intact, but only 58.5 per cent of the asset’s capital services remain in the exponential measure. At the other extreme, when the asset is 28 years old only 4.6 per cent of the gross stock, 3.4 per cent of the Denison stock, and 1 per cent of the net stock remains; however, 13.1 per cent of the asset is still in place under the exponential distribution. Hence, for an industry like Iron and Steel, where a large amount of investment took place in the 1950-1980 period, if the exponential distribution is used a significant amount of old, non-depreciated capital stock will still remain in place in the post-1980 period. Oulton and O’Mahony (1994, Table 3.4) provide the estimates of δ employed here. See also Fraumeni (1997) for an explanation of how DBR is combined with asset life to produce rates of deterioration/depreciation. 20 15 Appendix Table A2.1: Definitions of the 18 industries Indust Name SIC80 ry codes Asset class (and % distribution) A B C D E Iron and steel 221 – 223 13.8 3.7 77.9 Other metal 210, 224 1.4 10.0 20.0 56.5 manufacturing 3 Extraction, bricks, 231-239, 3.0 15.0 31.0 28.0 asbestos 241 – 246 4 Glass, pottery 247, 248 5.0 24.0 19.0 14.0 5 Chemicals 251 – 259 3.4 2.7 6.8 56.9 6 Man-made fibres 260 2.6 89.7 7 Other metal products 311 – 316 1.1 10.9 16.3 61.4 8 Electrical engineering 320 – 348 1.4 10.0 20.0 56.5 9 Motor vehicles 351 – 353 25.0 2.0 10.0 52.0 10 Shipbuilding 361 1.4 10.0 20.0 56.5 11 Other vehicles 362 – 365 3.0 13.0 69.0 12 Instrumental 371 – 374 1.4 10.0 20.0 56.5 Engineering 13 Food, drink, tobacco 411 – 429 2.0 22.0 68.0 14 Textiles 431 – 442 2.6 89.7 15 Clothing, footwear, 451 – 456 73.0 4.0 leather 16 Timber products 461 – 467 76.0 5.0 17 Paper, publishing 471 – 475 4.5 54.5 18 Other manufacturing 481 – 495 73.0 4.0 Average length of life for asset (in years): A=5; B=12; C=14; D=19; E=26; F=37. Source: ONS 1 2 F 4.6 12.1 23.0 38.0 30.2 7.7 10.3 12.1 11.0 12.1 15.0 12.1 8.0 7.7 23.0 19.0 41.0 23.0 16 3. Plant level Estimates of the Capital Stock 3.1 Gross investment data for plant and machinery was taken from the Annual Respondents Database (the ARD). The ARD basically comprises financial information21 collected from some 14-19,000 manufacturing establishments (or reporting units – denoted RU hereafter) for each year from 1973-200122, based on a stratified sampling frame that is heavily biased towards the largest establishments. Establishments (and the plants comprising such establishments) can be linked through time to form a panel, and information is also available on the population of establishments (or plants), which can be used to weight the financial data to obtain population estimates. 3.2 For each year there are two files that can be merged to produce plant level data; one file covers the sample of RU’s23, known as the ‘selected’ file, that were asked questions about financial matters (e.g. amounts spent on capital expenditure, including any pre-production expenditure), while the other contains information (such as employment and ownership structure) on ‘nonselected’ establishments (the remainder of the population). Establishment level data can be 'spread back' to local units (hereafter referred to as LU’s) using employment shares24 and the unique reference number allocated to each plant 21 Such as sales, purchases of inputs, investment undertaken, as well as the characteristics of respondents in terms of ownership and location. See footnote 1 for details. 22 Since 1998, data is available on an economy-wide basis and not just for production industries. However, since the perpetual inventory method used here to calculate capital stocks requires information on gross investment for up to 37 years, the economy-wide data cannot yet be used to calculate reliable plant level capital stock estimates. 23 Establishments are either single plants or they make a return that covers several plants - details and definitions are provided in the introductory notes for each annual census. 24 Note, employment data for local units is based on information supplied by respondents when they make a return to the ABI. Tit is this information on local unit employment that is used here. Employment data for non-selected local units is obtained from the IDBR (and before 1994 from the register of plants keep by the ONS), and for especially smaller plants this information may be interpolated or be based on information collected in previous years (thus making it less accurate). The 17 during the whole 1970-2001 period (with single unit establishments supplying their own LU – i.e. plant level - information). Sample data can be grossed-up to provide population estimates (using employment estimates for all plants)25 and each non-selected plant's share of annual investment in each industry can be apportioned back to these plants using employment share data. 3.3 Thus, plant-level (LU) estimates of capital expenditure are to some extent derived from establishment level (RU) information. The next chapter provides an analysis of the extent to which plants obtain their financial information through pro rata allocations from the RU to which they belong. In particular, large multi-plant establishments combine a significant proportion of their plants into single financial returns to the ABI. The extent to which plant level information on capital expenditure is biased cannot be measured, and especially whether this biases the overall estimates. Of course, ‘spreading back’ RU investment data to LU’s based on employment shares is to assume a constant investment-labour ratio across the local units of an establishment, when investment is usually ‘lumpy’ and for some plants is likely to involve ratios above and below the average for the RU. Over a number of years, however, this issue is unlikely to be too great a problem as there is a strong positive correlation between investment and employment levels across plants, reflecting the fact that large plants in employment terms on average tend to invest more (and vice versa). 3.4 In any case, using RU’s as the base for analysis results in a different set of issues which suggest that it is questionable whether using RU data is a viable source of employment data for non-selected plants (and thus weaknesses associated with its use) is discussed in the next Chapter. 25 The 'weights' were calculated at the 4-digit industry level and by RU size-band. Note, the 1980 Standard Industrial Classification was used throughout (requiring a ‘look-up’ table that reclassifies the 1992 SIC to the 1980 SIC), with plants from 1970-1979 reclassified from the 1968 SIC. 18 alternative to using plant-level data. As shown in Chapter 4, the main reason is that the establishments are not 'stable' over time (an essential requirement for the perpetual inventory measure of capital), since their composition (in terms of the number of plants they cover) can change as companies open and close plants, buy and sell plants, or simply change the way they report to the ONS (i.e., reallocate plants into different establishments for accounting purposes). Even those RU’s that have the same number of plants in different years (typically the smallest establishments), the year of opening and closing for the different plants in each RU are a mix of different dates. The largest RU’s, in employment terms, are more likely to have different numbers of plants in different years. The deviation in the opening and closing dates of the plants contained in these establishments is significant. In addition, the change in average employment in these establishments (and the standard deviation associated with the employment in the plants contained in each establishment) is very large. 3.5 Using plant-level estimates of capital expenditure (on plant and machinery), based on acquisitions less disposals and including pre-production expenditure (where available), it is possible to estimate the capital stock for each plant using the perpetual inventory method based on the Denison approach to deterioration and the length-of-life assumptions as used by the ONS when calculating the 'official' estimates for the UK (Chapter 2 provided details on the methods employed). For plants in existence in 1970, these estimates do not take account of pre-1970 investment. Thus, 1949-69 data supplied by the NIESR (based on 3-digit industries under the 1968 SIC) were used to calculate a 1969 benchmark series that was then apportioned to the 1970 stock of plants using 1970-72 19 shares of employment and gross expenditure on plant and machinery. The benchmark figures depreciate with no further new investment being added, and these estimates are then merged with the post-1969 capital stocks calculated for each plant. Plants in existence in 1970 that close after this date remove a proportion of the benchmark capital stock from the aggregate.26,27 3.6 Plant and machinery price deflators (based on the 1980 SIC) supplied by the ONS (and updated using information published annual in the MM14 series)28 were applied to the data, to produce real gross investment in plant and machinery by industry. The most disaggregated industry breakdown available which was consistent through time was used for 21 manufacturing sectors. The 1949-69 data supplied by the NIESR were in 1980 constant-prices. Updating the Capital Stock Estimates post-2001 3.7 The rest of this chapter explains in detail how to update the plant and machinery capital stock estimates produced here when ARD information for 2002 and 26 Specifically, the capital stocks of plants that closed were simply subtracted from the unadjusted industry stocks on a year-by-year basis from the date when the plant closed. The method used to measure the aggregate industry capital stock and individual plants capital stock is identical, being based on the same length of asset life assumptions and the same deterioration patterns. 27 Note, if a plant that closes belongs to a multi-plant enterprise, it is possible that some at least of its assets are distributed around the other plants in the establishment to which the plant belongs, without disposals and acquisitions being recorded in the financial returns for the establishment. To the (unknown) extent to which this happens, our estimates of scrapping will be biased upwards. However, it is likely that the majority of capital is ‘sunk’ in the sense of having only a specific use in the plant for which it was acquired. 28 Latest published figures are provided for the 1992 SIC, and these are matched as accurately as possible to the 1980 SIC. 20 beyond becomes available. The SPSS code used to do this is attached comprising four syntax files.29 3.8 The first step comprises merging the selected and non-selected files and calculating weights to obtain a single plant-level ARD file for the year in question. The syntax file that does this is labelled ‘2001 matching data.sps’, and is attached at the end of this report in the Statistical Appendix. The individual RU files for each sector (labelled dat2001xxx.sav or nul2001xxx.sav)30 are amalgamated and checked to ensure that there are no duplicate cases. These are then spread back to the LU data (after it has been merged into one file – each file is labelled snul2001xxx.sav). At this stage turnover, gross output, GVA, and gross investment in plant and machinery figures are obtained based on ONS definitions. Such estimates for LU data derived from RU data are multiplied by ‘unitwt’ which is the share of LU employment in total RU employment. 3.9 Step two comprises maintaining a plant level link through time (labelled CSO_REF), converting 1992SIC codes back to 1980SIC codes, and calculating real gross investment figures for all plants (whether they were ‘selected’ or ‘non-selected’). The syntax file for this is labelled ‘setup K 2001.sps’. The outcome is an updated file (labelled rnet_pm70xx.sav) which will be used to calculate capital stocks for each plant based on post-1969 real gross investment data. 3.10 Next, capital stock estimates based on the Denison approach are calculated for each plant. The syntax file is labelled ‘capstock.sps’. 29 The BDL unit at the ONS creates data files using STATA but these can be converted to SPSS files using the STAT TRANSFER software available at the BDL. Note, users can amend the programmes provided to meet their own needs if, for example, they decided to use a different depreciation series. 30 The xxx refer to the different industry codes (e.g. ‘prg’ refer to production) while the ‘dat’ refers to RU’s that were selected and sent forms for completing with regard to supplying financial information. The ‘nul’ files refer to non-selected RU’s that were not selected for the ABI. 21 Table 3.1: UK manufacturing plant & machinery capital stock 1970-2001 (£m 1980 prices) Year Capital Stock Real Gross Investment No. of Plants 1970 61,419 5,763 99410 1971 64,106 5,241 99021 1972 66,303 4,596 99631 1973 68,770 5,038 112063 1974 72,276 6,348 119848 1975 74,440 5,531 122113 1976 76,012 5,169 125189 1977 77,837 5,480 125952 1978 80,141 6,102 125957 1979 83,337 7,023 124703 1980 82,505 4,945 124032 1981 80,738 3,829 123242 1982 79,000 3,723 116659 1983 77,768 4,017 115498 1984 74,453 4,769 148925 1985 74,552 5,296 155719 1986 74,706 5,081 161600 1987 75,632 5,843 166109 1988 78,154 7,189 170702 1989 80,393 8,084 173933 1990 82,105 7,532 167208 1991 82,430 6,589 162532 1992 82,060 6,015 173144 1993 78,179 5,016 159526 1994 77,297 6,445 183453 1995 77,554 8,223 211276 1996 72,099 7,195 190946 1997 69,478 8,070 198338 1998 66,813 8,322 182726 1999 63,666 7,694 191308 2000 66,186 7,888 189135 2001 66,831 8,525 181653 Total 2,376,576 195,634 4798652 22 3.11 Finally, the capital stock data for individual plants is merged to pre-1970 benchmark data, to obtain the final estimates for each plant. The syntax file is labelled ‘capital stock final merge.sps’. The results for the 1970 to 2001 period for all manufacturing plants are presented in Table 3.1 23 [this page is blank] 24 4. Issues in Measuring the Capital Stock 4.1 A list of issues that need to be considered include the following: a. Whether plant (LU) or establishment (RU) level data should be used b. Implementation of the Perpetual Inventory method and specifically: i. What depreciation series should be used? ii. What start values are needed (i.e. the first year of gross investment data used) in order to derive an accurate series iii. The availability of plant level gross investment data (acquisitions net of disposals) iv. How to take account of plant closures 4.2 These issues are considered mainly by contrasting the approach set out in Chapters 2 and 3 with the approach used by Martin (2003).31 Table 4.1: RU Plant & Machinery Capital Stocks, 1980-1993, based on using LU and RU data Ratio of RU capital stocks LU data RU data £m 1980 prices % < 0.75 112,238 12.8 53,100 6.4 18,735 0.75 to <1 263,394 30.1 248,435 30.0 263,611 1 158,065 18.0 158,065 19.1 1,354,491 >1 to 1.25 319,201 36.4 331,845 40.1 271,688 > 1.25 22,929 2.6 36,312 4.4 11,744 Total 875,828 827,757 100 1,920,269 100 £m 1980 prices % No. of RU’s in each sub-group See text for details (author’s own calculations using the ARD) 31 Note we cannot compare directly our estimates with those generated by Martin as he calculates a measure including plant, machinery, buildings and vehicles and we only estimate a figure for plant and machinery. 25 Local versus reporting unit estimates of the capital stock 4.3 The arguments set out in par. 3.3 show that there are problems when using RU as opposed to LU data. The main reason is that RU’s are not 'stable' over time since their composition (in terms of the number of plants they cover) can change as companies open and close plants, buy and sell plants, or simply change the way they report to the ONS (i.e., reallocate plants into different establishments for accounting purposes). 4.4 In order to measure the implications of using RU’s (vis-à-vis LU’s) when estimating the capital stock, the 1970-93 real investment data for plant and machinery was re-assigned to RU’s. Then using the same procedures as in Chapter 3 but with RU reference codes rather than LU codes, the capital stock estimates were re-computed (referred to hereafter as the RU capital stock estimates). Note, pre-1970 benchmark data at RU level was not available, and thus for the present exercise only data from the ARD is used (covering plants and establishments from 1970 onwards). Also, RU codes that have been linked through time are not available post 1993 (when the ONS moved over to using new IDBR codes for LU’s and RU’s), since the ONS did not provide a complete look-up table to match local and reporting unit reference numbers in 1993 (the last time the VAT register was used) with those used in 1994 (when the IDBR was first used).32 4.5 Table 4.1 presents the results from aggregating plant level capital stock data to the RU level and comparing the results with the RU capital stock estimates. 32 The present author has fully linked LU’s using the old CSO_REF codes to maintain a link through time, but not RU’s, hence CSO_REF2 codes end in 1993. 26 Note, the extent to which the two series differ reflects the points noted in par. 4.3 that RU’s are often not stable over time (of course single plant enterprises have de facto the same LU and RU reference codes, and therefore will have identical capital stock estimates33). The first difference to note is that overall the total UK manufacturing capital stock is lower when RU data is used, since a number of RU’s were ‘closed’ during the period 1970-93 with (some or all) of the plants they accounted for reassigned to other RU’s. The consequence of this is that while the RU’s that gained these plants had higher levels of real investment after ‘acquiring’ the LU’s, they did not gain the accumulated past investments (i.e. the capital stocks) of such LU’s. 4.6 To recap, for each RU covering the 1980-1993 period included in Table 4.1, two estimates of capital stock were available: (i) one based on using plant level data aggregated to the RU level; and (ii) the other based on RU data. The difference between the two for each RU was calculated as the ratio of the estimate based on RU data to the estimate based on using LU data. During the 1980-93 period, over 70 per cent of RU’s were ‘stable’ in that they comprised the same plants throughout the post-1969 period (indeed, most were single plant enterprises – see the last column in Table 4.1), but these ‘stable’ RU’s only accounted for some 18 – 19 per cent of the total real capital stock during 198093. 4.7 For some RU’s, the capital stock based on using LU data was larger (so that the ratio is below 1 in Table 4.1). This was usually the result of RU’s that were closed having their plants reassigned but not the capital stocks of these reassigned plants (see par. 4.5 above). 33 Although this changes if they are acquired by a multi-plant firm. 27 Table 4.2: Plant & machinery capitals stock (1973=100) for single hypothetical RU, 1973-93 Year Total capital stock (K) No. of plants K lost due to ‘closure’ K gained due to plant reassigned from different RU 1973 100.0 30 1974 102.5 30 1975 105.1 30 1976 107.7 30 1977 110.4 30 1978 113.1 30 1979 116.0 30 1980 116.0 30 1981 77.3 20 1982 77.3 20 1983 77.3 20 1984 58.0 15 1985 58.0 15 1986 59.1 15 1987 59.1 15 1988 60.3 15 1989 57.6 15 1990 60.6 18 1991 72.7 18 1992 24.2 5 1993 24.2 5 Source: ARD and author’s own calculations 4.8 38.7 19.3 12.1 48.5 For other RU’s, the capital stock based on using LU data is smaller (ratios above 1 in Table 4.1). This typically resulted when some (but not all) of the plants within a RU were closed, which reduces the capital stock estimate based on using LU data but not the estimated obtained when using RU data (since the 28 RU does not close and thus none of the capital stock of closing plants is removed). 4.9 Table 4.2 makes this clearer; information for a single (hypothetical34) reporting unit is shown with capital stock data based on amalgamating the local units included in the RU. Thus, capital stock estimates here are based on using LU data. At the end of 1980, 1983 and 1991 several plants are ‘closed’35 (see the column headed ‘number of plants’), and consequently the total capital stock for the RU declines substantially. If the capital stock had been calculated using RU data, none of this reduction due to ‘closure’ would have taken place, leading to a significant overestimate of the actual capital stock available to this RU. 4.10 If a user of the ARD prefers to use RU data, then aggregating up LU estimates to the RU level will at least avoid the problems of ‘stability’ discussed in this section (although there still remains an issue with plant closures). 4.11 This sub-section concludes with a discussion of some of the problems associated with using plant level estimates of capital stock that are based on ‘spreading back’ gross investment data from RU’s to LU’s based on employment in local units. 4.12 The issue of ‘spreading back’ gross investment data from RU’s to LU’s, and thus imposing the assumption that investment-labour ratios are constant across the LU’s that make up a single RU, has already been discussed (par. 3.3). A further issue (and potential source of bias) is that employment data for nonselected plants is not always up-to-date or directly collected.36 34 These numbers are ‘made-up’ and do not relate to any actual RU to ensure confidentiality. Either closed in the sense of the plant(s) ceased production or the plant(s) were reassigned to another RU (or RU’s). 36 Of course, this is also true for non-selected RU’s (e.g. single-plant enterprises). Note also, any microbased analysis of the data will used (weighted) data for ‘selected’ plants, where employment information is up-to-date. 35 29 4.13 Since 1994, employment information is collected at local unit level by the Annual Register Inquiry (ARI), with larger firms (with over 50 employees) being covered annually and all firms updated at least every four years, from a variety of sources including ARI, ABI and the PAYE register. However, for very small firms where there is only VAT turnover information, employment is interpolated (based on the turnover/employment ratios of plants that do provide information on both measures). Moreover, smaller (usually single-plant) firms will have their employment information updated less frequently than larger firms, and therefore the employment information available for non-selected plants (belonging to non-selected RU’s) is less reliable in the short-term. 4.14 Prior to 1994, employment data for non-selected RU’s in the 0-19 employment size-band was based on interpolations using VAT turnover data and turnover/employment ratios from the 20-49 employment size-band. 4.15 This issue of the accuracy of using plant-level employment (and the unknown bias that may result) when ‘spreading-back’ data from RU’s needs to be weighed against the problems of using RU’s when calculating capital stock estimates (principally issues of the non-stability of RU’s and plant closures). Issues relating to implementation of the Perpetual Inventory method 4.16 The first issue to be considered is the depreciation series that is used in the present approach (based on Denison, 1967) compared to that used by Martin (2003). The latter uses a geometric depreciation rate based on the average ONS length of life for all manufacturing and thus δ = 0.06 in Martin (op. cit.) – see par. 2.12 for further details. Note, Martin does not use the length of life 30 information for each of the sub-groups within the 18 manufacturing industries for which data is available (see Table A2.1). Figure 2.1 (in Chapter 2) and the discussion in par. 2.12 and 2.13 illustrate the implications of using a geometric depreciation rate. It was argued that using such a depreciation rate removes capital services from the asset stock at too fast a rate in the initial years and then too slowly when the asset nears the end of its length of life. If we are interested in the economic services available from an asset, then the Denison (1972) approach produces a more appropriate deterioration patter (at first slow, followed by more rapid deterioration). 4.17 The next issue is the start values that are used in the perpetual inventory approach. In theory, we need gross investment data that extends back some 37 years before the first year for which we wish to calculate the plant and machinery capital stock.37 Plant level data on real gross investment in plant and machinery is only available from 1970, and this is linked to 3-digit industry level estimates of gross investment that covers the 1948-1969 period (see par 3.4). Investment data at the 3-digit level is not available before 1948, although more aggregated information can be obtained from the ONS (however, this was not used since it is felt that this would not be accurate enough for linking to plant-level data, and in any event much of the pre-1948 data covers the period of WWII and much of this stock of capital is likely to subject to special factors that make its use problematic38). Thus, the 1970-2001 plant level estimates of the (plant and machinery) capital stock comprise two (merged) elements: (i) estimates for each plant using the perpetual inventory method based on the 37 For estimates of the stock of industrial buildings, we would need gross investment data for at least 80 years before the first year for which we wish to estimate the capital stock, given the much longer length of life of buildings. 38 E.g. some of this stock would have been destroyed in the war; much would have been obsolete after the war and would have needed to be converted to peace-time use. 31 Denison approach to deterioration; and (ii) a 1969 benchmark series (based on the 1948-1969 3-digit 1968 SIC investment data) that was then apportioned to the 1970 stock of plants using 1970-72 shares of employment and gross expenditure on plant and machinery. The benchmark figures depreciate with no further new investment being added, and these estimates are then merged with the post-1969 capital stocks calculated for each plant. Plants in existence in 1970 that close after this date remove a proportion of the benchmark capital stock from the aggregate (further details are provided in par. 3.4). 4.18 Figure 4.1 shows what proportion of the total manufacturing capital stock (in selected years) is accounted for by the (depreciated) 1969 benchmark data. In the early years, plants that were open in 1970 dominate the aggregate stock and in turn the benchmark series provides the largest share of capital stock for such plants. By 1980, some 40 per cent of the aggregate stock is accounted for by investment that occurred pre-1970 (and thus which is based on the 3-digit 1968 SIC industry aggregate investment data). 4.19 In contrast, Martin (op. cit.) uses 2-digit (rather than 3-digit) SIC real gross investment data for 1948-1979 to calculate a 1979 benchmark series which is then apportioned to 1980 RU’s using the average material usage over the lifetime of a post-1979 RU. 4.20 The first difference to note when comparing the approach used here and Martin’s is that his capital stock figures for the 1980’s will be dominated by the benchmark series, as he only uses RU data from 1980 onwards. 4.21 Secondly, his method of apportioning the benchmark series to RU’s existing in 1980 is very different to the approach used in this study. It is unclear why apportioning using intermediate inputs (rather than investment or employment 32 data) is preferred; it is even harder to understand why the average over the lifetime of the RU should be used. For example, RU’s that are new in 1980 but survive a long time and grow relatively large will be apportioned a large part of the 1979 benchmark series in 1980, but RU’s that are relatively ‘old’ in 1980 and perhaps decline in size and close earlier will be allocated a smaller allocation of the 1979 benchmark series. Figure 4.1 Proportion of UK Plant & Machinery Capital stock in selected years accounted for by (depreciated) 1969 benchmark data 100 % 90 80 70 60 50 40 30 20 10 19 70 19 71 19 72 19 73 19 74 19 75 19 76 19 77 19 78 19 79 19 80 19 81 19 82 19 83 19 84 19 85 19 86 19 87 19 88 19 89 19 90 19 91 19 92 19 93 19 94 19 95 19 96 19 97 19 98 19 99 20 00 20 01 0 4.22 The approach used in this study apportioned the 1969 benchmark capital stock using the 1970-72 shares of investment and employment of 1970 plants (within each 3-digit industry), giving equal weight to the investment and employment shares.39 39 The problem with using investment shares is that investment (even covering a 3 year period) can be ‘lumpy’; the problem with using employment shares to proxy for the relative size of the plants within each 3-digit SIC industry is that this assumes a fixed ratio between the size of the capital stock and the size of the employment stock in 1969. So both series were used, although it was found that the two were strongly correlated across plants suggesting that using either series would not be inappropriate. 33 4.23 The third issue associated with implementing the perpetual inventory method is that it requires investment data (acquisitions net of disposals)40 in every year that the plant (or RU) is open, not just those years when the unit was selected to be included in the ABI (or Annual Census of Production). In this study, data for ‘non-selected’ years was interpolated by grossing-up sample data (provided in the ‘selected’ file) to provide population estimates of gross investment for each 4-digit industry (using the 1980 SIC)41, subtracting the share of investment attributable to the ‘selected’ plants so that what is left over comprises the nonselected share of annual investment in each industry, which is then apportioned back to non-selected plants using employment share data. Figure 4.2 provides an example of the procedure for one plant covering the 1973 – 1987 period (the red diamonds denote the interpolated investment levels in each year). 4.24 In contrast, Martin (op. cit.) linearly interpolates between any two years where selected data for an RU is available (and if data is missing prior to closure then the last observed value is used). In terms of Figure 4.2, the equivalent estimates using Martin’s approach would lie on the blue dashed lines in the diagram. Clearly a major weakness of Martin’s approach is that it takes no account of the level of total investment that took place in any year, since adding together his estimates of selected and non-selected (interpolated) investment data is not constrained to equal the total level of (population weighted) investment for an industry. That is why in Figure 4.2 the interpolated data used here lies above and below the estimates that would be obtained using Martin’s technique; 40 Note, from 1976 data on leasing of plant and machinery is available separately in the ARD. These figures were not used here, but should added to the capital stock (after deflation) prior to use in any econometric analysis (such as estimating TFP). 41 This is the procedure used by the ONS when publishing industry level investment data. 34 fluctuations are included here to reflect overall investment levels in each industry in each year. 4.25 The final issue is how to take account of closures. The section above dealing with the implications of using LU versus RU data (par. 4.3 to 4.14) shows that when firms reduce their production capacity (through closing a plant) this cannot be adequately dealt with if RU data is used. Estimates of the capital stock using RU data will therefore be biased when closures occur, and the extent of bias is likely to be relatively large. Figure 4.2 Actual and interpolated/extrapolated gross investment for one plant in UK manufacturing, 1973-1987 (1973=1) 3.5 3.0 interpolated data (Harris approach) 2.5 2.0 1.5 1.0 interpolated data (Martin) 0.5 0.0 1973 1974 1975 1976 1977 1978 1979 1980 1981 1982 1983 1984 1985 1986 1987 -0.5 -1.0 -1.5 35 Conclusions and recommendations 4.26 To conclude, the methods used here to calculate the plant and machinery capital stock are argued to have certain advantages over those employed by Martin (and others using RU data) since they: • are based on LU data and thus avoid the problem that LU’s can be reallocated between RU’s for accounting purposes by firms, or when plants are bought and sold, or when they are opened and closed; • the Denison approach to calculating economic deterioration is argued to be more realistic when compared to using a geometric depreciation series, especially if the purpose of calculating capital stock estimates is to use these when estimating a production function or when estimating total factor productivity; • post-1969 data at the plant level is used here, while Martin uses post-1979 data, and thus the influence of benchmark data is lower in the present estimates, especially from 1980 onwards. The benchmark series used here were also more disaggregated, and thus likely to have been more accurately distributed to those plants that were open in 1970 (and thus required pre-1970 capital stocks to be allocated to them), especially as Martin uses the average value of intermediate inputs of RU’s based on their entire operating life to allocate the benchmark data (rather than investment and employment shares linked to the benchmark period); • investment data is interpolated (when necessary) using a population estimate of investment in a 4-digit industry in any particular year (after subtracting investment already identified to belong to those plants that 36 were selected for inclusion in the ABI), while Martin simply interpolates linearly between observed investment levels for each RU; • the capital stock produced here takes account of plant closures, while Martin does not. 4.27 However, ‘spreading back’ investment data from selected RU’s based on LU employment shares does introduce an unknown level of bias into the capital stock estimates produced at the plant level. Improvements in the accuracy of measuring the capital stock using the perpetual inventory approach would be achieved if the ONS were to consider the following changes42: • collecting information on capital expenditure at the local unit level (in addition to employment data) when conducting the ABI (this information was apparently collected prior to 1994 when RU’s were asked to report on employment and capital expenditure for the plants included in the return, but the information was not made available in the ARD); • increasing the frequency with which LU employment estimates are updated for smaller firms. The National Assembly for Wales has for some years funded a top-up to the ARI for the region and Scotland has recently agreed the same with the ONS. The additional costs are not large and it is estimated that the benefits in terms of the improved accuracy of employment and plant turnover information for smaller firms (below 50 employees) will be significant.43 42 The cost of such changes is not considered here and the ONS would need to balance costs against gains in terms of the quality of the micro-data generated, given that the primary purpose of the ABI is to generate information which is used to provide figures at a much more aggregated level (e.g. national GVA statistics by industry). However, updating the IDBR through the Annual register Inquiry is also important since the IDBR underpins analysis of business demography at the local level, which is important to government in terms of policy aims and objectives. 43 The Scottish contact person is Gerhard Mors ([email protected]) who has further information. 37 • obtaining more up-to-date estimates of economic depreciation of assets by industry, in order to provide better estimates of the rate of deterioration and obsolescence. 4.28 An alternative might be for ONS to undertake a limited survey of plants and firms in order to generate direct estimates of the capital stock. This would not only confirm how accurate the methods used here are (especially in relative terms across plants), but would also help to provide an alternative to the perpetual inventory approach which comprises the standard means for calculating capital stock in most countries. 38 References Anderson, W. P. and D. L. Rigby (1989) Estimating Capital Stocks and Capital Ages in Canada’s Regions: 1961-1981, Regional Studies, 23, 117-126. Bahk, B.H., and M. Gort (1993) Decomposing Learning by Doing in New Plants, Journal of Political Economy, 101, 561-83. Barna, T. (1962) Investment and Growth Policies in British Industrial Firms, Occasional Paper No. 20, National Institute of Economic and Social Research, Cambridge University Press, Cambridge. Bartelsman, E.J., and P.J. Dhrymes (1998) Productivity Dynamics: U.S. Manufacturing Plants, 1972-1986, Journal of Productivity Analysis, 9, 5-34. Coen, R.M. (1975) Investment Behaviour, the Measurement of Depreciation and Tax Policy, American Economic Review, 65, 59-75. Coen, R. M. (1980) Alternative Measures of Capital and its Rate of Return in United States Manufacturing, in Usher D (ed.) The Measurement of Capital, 121-48, National Bureau of Economic Research, Studies in Income and Wealth, Vol. 45, University of Chicago Press, Chicago. Denison, E. F. (1972) Final Comments, Survey of Current Business, 52, 95-110. Disney, R., J. Haskel and Y. Heden, (2003a) Entry, exit and establishment survival in UK manufacturing, Journal of Industrial Economics, 51, 93-115. Disney, R., J. Haskel and Y. Heden, (2003b) Restructuring and Productivity Growth in UK Manufacturing, Economic Journal, 2003, 113, 666-694. Doms, M. ,T. Dunne and M. J. Roberts, (1995) The Role of Technology use in the Survival and Growth of Manufacturing Industries, Rand Journal of Economics, 19, 495-515. Doms, M.E. (1996) Estimating Capital Efficiency Schedules within Production Functions, Economic Inquiry; 34, 78-92. Fabricant, S. (1980) Comment, in Usher D (ed.) The Measurement of Capital, 149-52, National Bureau of Economic Research, Studies in Income and Wealth, Vol. 45, University of Chicago Press, Chicago. Fraumeni, B.M. (1997) Measurement of Depreciation in the U.S. National Income and Wealth Accounts, Survey of Current Business, 74, 7-23. 39 Gorzig, B. and A. Stephan (2002) Outsourcing and Firm-level Performance, German Institute for Economic Research, Discussion paper 309, DIW Berlin. Griffith, R (1999) Using the ARD Establishment Level Data to Look at Foreign Ownership and Productivity in the United Kingdom, the Economic Journal, 109, F416-F442. Harris, R. I. D. and P. Hassaszadeh (2002) The Impact of Ownership Changes and Age Effects on Plant Exits in UK Manufacturing, 1974-1995, Economics Letters, 75, 309-317. Harris, R. I. D. and S. Drinkwater (2000) UK Plant and Machinery Capital Stocks and Plant Closures, Oxford Bulletin of Economics and Statistics, 62, 239-261. Harris, R.I.D. (2001) Comparing Regional Technical Efficiency in UK Manufacturing Plants: The Case of Northern Ireland 1974-1995, Regional Studies, 35, 519354 Harris, R.I.D (2002) Foreign Ownership and Productivity in the United Kingdom Some Issues When Using the ARD Establishment Level Data, Scottish Journal of Political Economy, 49, 318-335 Harris, R.I.D and A. Collins (2002) Pollution Abatement Expenditure, Productive Efficiency and Plant Ownership, Land Economics, 78, 171-189 Harris, R.I.D and C. Robinson (2002) The Impact of Foreign Acquisitions on Total Factor Productivity: Plant-level Evidence from UK Manufacturing, 19871992, Review of Economics and Statistics, 84, 562-568 Harris, R.I.D and C. Robinson (2003) Foreign Ownership and Productivity in the United Kingdom: Estimates for UK Manufacturing Using the ARD, Review of Industrial Organisation, 22, 207-223 Harris, R.I.D and C. Robinson (2004a) Productivity Impacts and Spillovers from Foreign Ownership in the United Kingdom, National Institute Economic Review, 187, 70-87. Harris, R.I.D and C. Robinson (2004b) Industrial Policy and its Effect on Total Factor Productivity in UK Manufacturing Plants, 1990-1998, forthcoming Scottish Journal of Political Economy. Harris, R.I.D. (2004) DTI Industrial Support Policies: Key Findings from New Microdata Analysis, in Raising UK Productivity – Developing the Evidence Base for Policy, DTI Economics Paper No. 8, London. 40 Haskel, J. and R. Martin (2003), Productivity Spreads Using Conventional Measures, CeRiBA Discussion Paper ( http://www.CeRiBA.org.uk ) Hulten, C. R. (1992) Growth Accounting When Technical Change is Embodied in Capital, American Economic Review, 82, 964-980. Hulten, C. R. and F. C. Wykoff (1981) The Estimation of Economic Depreciation Using Vintage Asset Prices: An Application of the Box-Cox Transformation, Journal of Econometrics, 15, 367-396. Hulten, C. R. and F. C. Wykoff (1996) Issues in the Measurement of Economic Depreciation: Introductory Remarks, Economic Inquiry, 34, 10-23. Jefferson, C. W. (1971) Capital Statistics for Irish Manufacturing Industry, Paper No. 60, Economic and Social Research Institute, Dublin. Kalt, J. P. (1978) Technological Change and Factor Substitution in the United States, 1929-67, International Economic Review, 19, 761-775. Kovenock, D. and G.M. Phillips (1997) Capital Structure and Product Market Behaviour: An Examination of Plant Exit and Investment Decisions, The Review of Financial Studies, 10, 767-803. Martin, R (2003) Building the capital stock, CeRiBA (version dated 11/2/03). McGuckin, R.H. and S.V. Nguyen (2001) The Impact of Ownership Changes: A View from Labor Markets, International Journal of Industrial Organization, 19, 739-762. Nadiri, M. and I. R. Prucha (1996) Estimation of the Depreciation Rate of Physical and R&D Capital in the U.S. Total Manufacturing Sector, Economic Inquiry, 24, 43-56. Olley, G. S. and A. Pakes (1996) The Dynamics of Productivity in the Telecommunications Equipment Industry, Econometrica, 64, 1263-97. Oulton, N. and M. O’Mahony (1994) Productivity and Growth: A Study of British Industry, 1954-1986, The National Institute of Economic and Social research Occasional Papers 47, Cambridge University Press, Cambridge. Oulton, N. (1997) The ABI Respondents Database: A New Resource for Industrial Economics Research, Economic Trends, 528, 46-57. Triplett, J.E. (1996) Depreciation in Production Analysis and in Income Wealth Accounts: Resolution of an Old Debate, Economic Inquiry, 34, 93-115. 41 Statistical Appendix 2001 matching data.sps * MERGE RU'S FOR DIFFERENT SECTORS. set mxmemory=2097151. get file= 'h:\files\dat2001cag.sav' . add files file=* /file='h:\files\dat2001cng.sav' /file='h:\files\dat2001mtg.sav' /file='h:\files\dat2001pdg.sav' /file='h:\files\dat2001prg.sav' /file='h:\files\dat2001reg.sav' /file='h:\files\dat2001stg.sav' /file='h:\files\dat2001whg.sav' /file='h:\files\nul2001cag.sav' /file='h:\files\nul2001cng.sav' /file='h:\files\nul2001mtg.sav' /file='h:\files\nul2001pdg.sav' /file='h:\files\nul2001prg.sav' /file='h:\files\nul2001reg.sav' /file='h:\files\nul2001stg.sav' /file='h:\files\nul2001whg.sav' . * dlink_re REFERS TO RU NUMBER. sort cases by dlink_re. compute n=1. * CHECK FOR DUPLICATE CASES. aggregate outfile='h:\files\del.sav' /break=dlink_re/number=sum(n). match files file=*/table='h:\files\del.sav'/by dlink_re. frequencies vars=number. compute ru_emp=empment. save outfile='h:\files\ru2001gb.sav'. sort files by number dlink_re. * MERGE LU DATA FOR DIFFERENT SECTORS. get file= 'h:\files\snul2001cag.sav' . add files file=* /file='h:\files\snul2001cng.sav' /file='h:\files\snul2001mtg.sav' /file='h:\files\snul2001pdg.sav' /file='h:\files\snul2001prg.sav' /file='h:\files\snul2001reg.sav' /file='h:\files\snul2001stg.sav' /file='h:\files\snul2001whg.sav' . sort cases by dlink_re. save outfile='h:\files\snul2001gb.sav'. descriptives dlink_r1. * MERGE RU DATA BACK TO LU DATA. get file='h:\files\snul2001gb.sav'. match files file=*/table='h:\files\ru2001gb.sav' /by dlink_re. execute. if (ru_emp gt 0) unitwt=empment/ru_emp. if (empment eq 0) unitwt=1. if (missing(dlink_r1)) dlink_r1=dlink_re. * RENAME LU AND RU CODES TO luref AND ruref. compute luref=dlink_r1. compute ruref=dlink_re. descriptives unitwt. descriptives wq11. save outfile='h:\files\capital stock\2001gb_all.sav'/drop=dlink_r1 dlink_re. * WEIGHT DATA. get file='h:\files\capital stock\2001gb_all.sav'. compute selected=0. if (wq11 gt 0) selected=1. *means unitwt by selected. compute number=1. compute sic92a=trunc(sic92/10). aggregate outfile='h:\files\weight1c.sav' /break=sic92a selected /pop4=sum(empment) /nsamp4=sum(number). aggregate outfile='h:\files\weight2c.sav' 42 /break=sic92a /totpop4=sum(empment). sort cases by sic92a selected. match files file=*/table='h:\files\weight1c.sav'/by sic92a selected. match files file=*/table='h:\files\weight2c.sav'/by sic92a . compute totemp1=empment. do if (selected eq 1 ). compute nsamp=nsamp4. compute popwt=totpop4/pop4. end if. recode nsamp (1 thru 5=1) (6 thru 10=2) (11 thru 20=3) (21 thru high=4) into nsamp5. *crosstabs sic92a by nsamp5. temporary. select if selected eq 1. descriptives empment /statistics=sum . weight by popwt. temporary. select if selected eq 1. descriptives empment /statistics=sum . weight off. compute weight=popwt*(9621759/21623246). descriptives weight popwt. * COMPUTE TURNOVER etc DATA FOR EACH PLANT, HAVING MULTIPLIED BY unitwt. compute turnover=wq399*unitwt. if (missing(turnover)) turnover=wq346*unitwt. if (missing(wq11)) turnover=-99. missing values turnover (-99). compute go=turnover+((wq500wq599)*unitwt)+(wq602*unitwt). if (missing(go) and missing(wq602)) wq602=0. if (missing(go)) go=turnover+((wq501wq502)*unitwt)+(wq602*unitwt). if (missing(wq416)) wq416=0. compute gva=turnover+(wq317*unitwt)+((wq500 -wq599)*unitwt)+(wq602*unitwt)(wq499*unitwt)+(wq414*unitwt)(wq400*unitwt)+(wq416*unitwt). if (missing(gva)) gva=turnover+(wq317*unitwt)+((wq501 -wq502)*unitwt)+(wq602*unitwt)(wq499*unitwt)+(wq414*unitwt)(wq400*unitwt)+(wq416*unitwt). means wq11 gva turnover go gva by sic92a/cells=sum count. variable label gva'GVA at FC (x unitwt)' go'gross output (x unitwt)' turnover'turnover (x unitwt)'. * RELABEL luref TO idbr_ref. compute idbr_ref=luref. sort cases by idbr_ref. save outfile='h:\files\capital stock\2001gb_all.sav'/drop=nsamp4 nsamp5 totpop4 pop4 sic92a totemp1 number /compressed. * CALCULATE PLANT AND MACHINERY GROSS INVESTMENT. get file='h:\files\capital stock\2001gb_all.sav'. descriptives wq515 wq516 wq527 wq530. if missing(wq515) wq515=0. if missing(wq516) wq516=0. if missing(wq527) wq527=0. if missing(wq530) wq530=0. compute netpm=((wq527wq530)+(wq515-wq516))*unitwt. descriptives netpm. save outfile='h:\files\capital stock\2001gb_all.sav'. 43 setup K 2001.sps * STEP ONE IS TO LINK IDBR_REF TO CSO_REF USING MOST RECENT YEAR WHERE LINK EXISTS. get file='h:\files\capital stock\2000_merge.sav'/keep=cso_ref idbr_ref . aggregate outfile='h:\files\capital stock\idbr look up.sav' /break=idbr_ref /cso_ref=min(cso_ref). * STEP TWO IS TO LINK TO SIC80 AND CSO_REF, AND CREATE NEW CSO_REFS FOR PLANTS THAT OPENED THIS YEAR. GET FILE='h:\files\capital stock\2001gb_all.sav'. * SELECT JUST MANUFACTURING PLANTS. select if sic92 eq 74200 or (sic92 gt 14000 and sic92 lt 40000). execute. * MERGE IN DATA FOR MATCHING 1992SIC CODES TO 1980SIC CODES. sort cases by sic92. match files file=*/table='h:\files\capital stock\post-93 SIC80 agg codes.sav'/by sic92. compute idbr_ref=luref. * MATCH IN cso_ref LOOK-UP DATA FOR PREVIOUS YEAR. sort cases by idbr_ref. match files file=*/table='h:\files\capital stock\idbr look up.sav'/by idbr_ref. descriptives idbr_ref cso_ref. sort cases by cso_ref. * FOR PLANTS WITH NO CSO_REF, CREATE NEW NUMBERS STARTING FROM HIGHEST CSO_REF VALUE (denoted ?) ALREADY EXISTING. do if missing(cso_ref). compute cso_ref=$casenum+?. end if. save outfile='h:\files\capital stock\2001gb_merge.sav'. *STEP THREE: SAVE NET PLANT INVESTMENT EXPENDITURE INTO NEW FILE. get file='h:\files\capital stock\2001gb_merge.sav'. compute net_pm=netpm*1000. * IF not yet in production INVESTMENT DATA IS MISSING THAT ASSIGN zero TO IT. compute npm_nyip=0. means net_pm npm_nyip by sic80/cells=sum. save outfile='h:\files\capital stock\2001 net cap.sav'/keep=cso_ref year sic80 net_pm npm_nyip popwt empment. *STEP FOUR: COMPUTE INVESTMENT FIGURES FOR NONSELECTED PLANTS AND DEFLATE TO REAL PRICES. get file='h:\files\capital stock\2001 net cap.sav'. if (missing(popwt)) popwt=0. compute wnetpm=0. compute wnetpm=net_pm*popwt. compute emp=0. if (popwt eq 0) emp=empment. aggregate outfile='h:\files\capital stock\delete.sav' /break=year sic80 /totem=sum(emp) /totnpm=sum(wnetpm) /totanpm=sum(net_pm). sort cases by year sic80. match files /file=*/table='h:\files\capital stock\delete.sav'/by year sic80. execute. if (popwt eq 0) net_pm=(emp/totem)*(totnpm-totanpm). temporary. select if (sic80 ge 2000 and sic80 lt 5000). 44 means net_pm by year/cells=sum mean count. if missing(npm_nyip) npm_nyip=0. compute net_pm=net_pm+npm_nyip. sort cases by cso_ref year. temporary. select if (sic80 ge 2000 and sic80 lt 5000). means net_pm by year/cells=sum mean count. * RECODE SIC80 INTO 21 INDUSTRY GROUPS FOR DEFLATION LATER. compute ind=trunc(sic80/100). if (sic80 ge 4710 and sic80 le 4729) ind=471. if (sic80 ge 4750 and sic80 lt 4759) ind=475. recode ind (21=22) (26=25) . select if (sic80 gt 2000 and sic80 lt 5000). * GROUP INTO THE 18 INDUSTRIES USED FOR LENGTH OF LIFE DEPRECIATION. recode sic80 (2210 thru 2239=1) (2100 thru 2199=2) (2245 thru 2247=2) (2300 thru 2399=3) (2410 thru 2469=3) (2470 thru 2479=4) (2480 thru 2489=4) (2500 thru 2599=5) (2600=6) (3100 thru 3199=7) (3200 thru 3499=8) (3500 thru 3530=9) (3610=10) (3620 thru 3650=11) (3700 thru 3799=12) (4110 thru 4299=13) (4300 thru 4399=14) (4400 thru 4599=15) (4600 thru 4699=16) (4700 thru 4799=17) (4800 thru 4999=18) into ind1. frequencies vars=ind ind1. temporary. select if missing(ind1). frequencies vars=sic80. * DEFLATE INVESTMENT DATA. sort cases by year ind. match files file=*/table='h:\files\capital stock\capital deflators.sav'/by year ind. compute rnet_pm=net_pm/def. temporary. select if sic80 ge 2100 and sic80 lt 5000. means rnet_pm net_pm by year. save outfile='h:\files\capital stock\capex_01.sav'/keep=year cso_ref ind1 rnet_pm. * ADD 2001 DATA TO 1970-2000 REAL INVETMENT DATA. get file='h:\files\capital stock\rnet_pm_7000.sav'. add files file=*/file='h:\files\capital stock\capex_01.sav'. select if (rnet_pm gt -20000000 and rnet_pm lt 322000000). select if (cso_ref ge 0 and cso_ref lt 9900000000000). *descriptives all. frequencies vars=year. means rnet_pm by year/cells=sum. descriptives ind1 . * GROUP DATA INTO 22 SUBGROUPS AS FILE TOO BIG FOR SINGLE RUN. recode cso_ref (low thru 1397711=1) (1397712 thru 1909602=2) (1909603 thru 2367495=3) (2367496 thru 2726899=4) (2726900 thru 2962701=5) (2962702 thru 3188255=6) (3188256 thru 3373448=7) (3373449 thru 3615079=8) (3615080 thru 3849495=9) (3849496 thru 4066445=10) (4066446 thru 4530242=11) (4530243 thru 4630730=12) (4630731 thru 4659740=13) (4659741 thru 4702195=14) (4702196 thru 4765747=15) (4765745 thru 5004878=16) (5004879 thru 6096082=17) (6096083 thru 6237839=18) (6237840 thru 6268857=19) (6268858 thru 9000000=20) (9000000 thru 100000000=21) (100000000 thru high=22) into group. 45 frequencies vars=group . save outfile='h:\files\capital stock\rnet_pm_7001.sav'/keep=cso_ref year ind1 rnet_pm group. 46 capstock.sps. set mxmemory=2097151. get file='h:\files\capital stock\rnet_pm_7001.sav'. * RUN 22 TIMES FROM group eq 1 TO group eq 22 SINCE OTHERWISE TOO BIG TO COMPLETE IN FEWER RUNS. * REMEMBER TO CHANGE group eq x AND pmcx EACH RUN WHERE x=1 TO 22. select if group eq 1. execute. * MATCH IN LENGTH OF LIFE ASSUMPTIONS. sort cases by ind1. match files file=*/table='h:\files\capital stock\length of life.sav'/by ind1. execute. compute test=g1+g2+g3+g4+g5+g6. frequencies test. if (year eq 1993 and cso_ref gt 80000000) cso_ref=cso_ref-90000000. * THROUGHOUT IF DATA POST 2001 IS ADDED, NEW LINES NEED TO BE ADDED FOR ADDITONAL YEARS. if (year eq 1970) g1_70=g1*rnet_pm. if (year eq 1971) g1_71=g1*rnet_pm. if (year eq 1972) g1_72=g1*rnet_pm. if (year eq 1973) g1_73=g1*rnet_pm. if (year eq 1974) g1_74=g1*rnet_pm. if (year eq 1975) g1_75=g1*rnet_pm. if (year eq 1976) g1_76=g1*rnet_pm. if (year eq 1977) g1_77=g1*rnet_pm. if (year eq 1978) g1_78=g1*rnet_pm. if (year eq 1979) g1_79=g1*rnet_pm. if (year eq 1980) g1_80=g1*rnet_pm. if (year eq 1981) g1_81=g1*rnet_pm. if (year eq 1982) g1_82=g1*rnet_pm. if (year eq 1983) g1_83=g1*rnet_pm. if (year eq 1984) g1_84=g1*rnet_pm. if (year eq 1985) g1_85=g1*rnet_pm. if (year eq 1986) g1_86=g1*rnet_pm. if (year eq 1987) g1_87=g1*rnet_pm. if (year eq 1988) g1_88=g1*rnet_pm. if (year eq 1989) g1_89=g1*rnet_pm. if (year eq 1990) g1_90=g1*rnet_pm. if (year eq 1991) g1_91=g1*rnet_pm. if (year eq 1992) g1_92=g1*rnet_pm. if (year eq 1993) g1_93=g1*rnet_pm. if (year eq 1994) g1_94=g1*rnet_pm. if (year eq 1995) g1_95=g1*rnet_pm. if (year eq 1996) g1_96=g1*rnet_pm. if (year eq 1997) g1_97=g1*rnet_pm. if (year eq 1998) g1_98=g1*rnet_pm. if (year eq 1999) g1_99=g1*rnet_pm. if (year eq 2000) g1_00=g1*rnet_pm. if (year eq 2001) g1_01=g1*rnet_pm. * E.G. IF 2002 DATA IS BEING ADDED, ADD NEW LINE HERE AND REPEAT FOR ALL OTHER ENTRIES WHERE 2002 NEEDS TO BE ADDED. if (year eq 1970) g2_70=g2*rnet_pm. if (year eq 1971) g2_71=g2*rnet_pm. if (year eq 1972) g2_72=g2*rnet_pm. if (year eq 1973) g2_73=g2*rnet_pm. if (year eq 1974) g2_74=g2*rnet_pm. if (year eq 1975) g2_75=g2*rnet_pm. if (year eq 1976) g2_76=g2*rnet_pm. if (year eq 1977) g2_77=g2*rnet_pm. if (year eq 1978) g2_78=g2*rnet_pm. if (year eq 1979) g2_79=g2*rnet_pm. if (year eq 1980) g2_80=g2*rnet_pm. if (year eq 1981) g2_81=g2*rnet_pm. if (year eq 1982) g2_82=g2*rnet_pm. if (year eq 1983) g2_83=g2*rnet_pm. if (year eq 1984) g2_84=g2*rnet_pm. if (year eq 1985) g2_85=g2*rnet_pm. if (year eq 1986) g2_86=g2*rnet_pm. if (year eq 1987) g2_87=g2*rnet_pm. if (year eq 1988) g2_88=g2*rnet_pm. if (year eq 1989) g2_89=g2*rnet_pm. if (year eq 1990) g2_90=g2*rnet_pm. if (year eq 1991) g2_91=g2*rnet_pm. if (year eq 1992) g2_92=g2*rnet_pm. if (year eq 1993) g2_93=g2*rnet_pm. if (year eq 1994) g2_94=g2*rnet_pm. if (year eq 1995) g2_95=g2*rnet_pm. if (year eq 1996) g2_96=g2*rnet_pm. 47 if (year eq 1997) g2_97=g2*rnet_pm. if (year eq 1998) g2_98=g2*rnet_pm. if (year eq 1999) g2_99=g2*rnet_pm. if (year eq 2000) g2_00=g2*rnet_pm. if (year eq 2001) g2_01=g2*rnet_pm. if (year eq 1970) g3_70=g3*rnet_pm. if (year eq 1971) g3_71=g3*rnet_pm. if (year eq 1972) g3_72=g3*rnet_pm. if (year eq 1973) g3_73=g3*rnet_pm. if (year eq 1974) g3_74=g3*rnet_pm. if (year eq 1975) g3_75=g3*rnet_pm. if (year eq 1976) g3_76=g3*rnet_pm. if (year eq 1977) g3_77=g3*rnet_pm. if (year eq 1978) g3_78=g3*rnet_pm. if (year eq 1979) g3_79=g3*rnet_pm. if (year eq 1980) g3_80=g3*rnet_pm. if (year eq 1981) g3_81=g3*rnet_pm. if (year eq 1982) g3_82=g3*rnet_pm. if (year eq 1983) g3_83=g3*rnet_pm. if (year eq 1984) g3_84=g3*rnet_pm. if (year eq 1985) g3_85=g3*rnet_pm. if (year eq 1986) g3_86=g3*rnet_pm. if (year eq 1987) g3_87=g3*rnet_pm. if (year eq 1988) g3_88=g3*rnet_pm. if (year eq 1989) g3_89=g3*rnet_pm. if (year eq 1990) g3_90=g3*rnet_pm. if (year eq 1991) g3_91=g3*rnet_pm. if (year eq 1992) g3_92=g3*rnet_pm. if (year eq 1993) g3_93=g3*rnet_pm. if (year eq 1994) g3_94=g3*rnet_pm. if (year eq 1995) g3_95=g3*rnet_pm. if (year eq 1996) g3_96=g3*rnet_pm. if (year eq 1997) g3_97=g3*rnet_pm. if (year eq 1998) g3_98=g3*rnet_pm. if (year eq 1999) g3_99=g3*rnet_pm. if (year eq 2000) g3_00=g3*rnet_pm. if (year eq 2001) g3_01=g3*rnet_pm. if (year eq 1970) g4_70=g4*rnet_pm. if (year eq 1971) g4_71=g4*rnet_pm. if (year eq 1972) g4_72=g4*rnet_pm. if (year eq 1973) g4_73=g4*rnet_pm. if (year eq 1974) g4_74=g4*rnet_pm. if (year eq 1975) g4_75=g4*rnet_pm. if (year eq 1976) g4_76=g4*rnet_pm. if (year eq 1977) g4_77=g4*rnet_pm. if (year eq 1978) g4_78=g4*rnet_pm. if (year eq 1979) g4_79=g4*rnet_pm. if (year eq 1980) g4_80=g4*rnet_pm. if (year eq 1981) g4_81=g4*rnet_pm. if (year eq 1982) g4_82=g4*rnet_pm. if (year eq 1983) g4_83=g4*rnet_pm. if (year eq 1984) g4_84=g4*rnet_pm. if (year eq 1985) g4_85=g4*rnet_pm. if (year eq 1986) g4_86=g4*rnet_pm. if (year eq 1987) g4_87=g4*rnet_pm. if (year eq 1988) g4_88=g4*rnet_pm. if (year eq 1989) g4_89=g4*rnet_pm. if (year eq 1990) g4_90=g4*rnet_pm. if (year eq 1991) g4_91=g4*rnet_pm. if (year eq 1992) g4_92=g4*rnet_pm. if (year eq 1993) g4_93=g4*rnet_pm. if (year eq 1994) g4_94=g4*rnet_pm. if (year eq 1995) g4_95=g4*rnet_pm. if (year eq 1996) g4_96=g4*rnet_pm. if (year eq 1997) g4_97=g4*rnet_pm. if (year eq 1998) g4_98=g4*rnet_pm. if (year eq 1999) g4_99=g4*rnet_pm. if (year eq 2000) g4_00=g4*rnet_pm. if (year eq 2001) g4_01=g4*rnet_pm. if (year eq 1970) g5_70=g5*rnet_pm. if (year eq 1971) g5_71=g5*rnet_pm. if (year eq 1972) g5_72=g5*rnet_pm. if (year eq 1973) g5_73=g5*rnet_pm. if (year eq 1974) g5_74=g5*rnet_pm. if (year eq 1975) g5_75=g5*rnet_pm. if (year eq 1976) g5_76=g5*rnet_pm. if (year eq 1977) g5_77=g5*rnet_pm. if (year eq 1978) g5_78=g5*rnet_pm. if (year eq 1979) g5_79=g5*rnet_pm. if (year eq 1980) g5_80=g5*rnet_pm. if (year eq 1981) g5_81=g5*rnet_pm. if (year eq 1982) g5_82=g5*rnet_pm. if (year eq 1983) g5_83=g5*rnet_pm. if (year eq 1984) g5_84=g5*rnet_pm. if (year eq 1985) g5_85=g5*rnet_pm. if (year eq 1986) g5_86=g5*rnet_pm. if (year eq 1987) g5_87=g5*rnet_pm. if (year eq 1988) g5_88=g5*rnet_pm. if (year eq 1989) g5_89=g5*rnet_pm. 48 if (year eq 1990) g5_90=g5*rnet_pm. if (year eq 1991) g5_91=g5*rnet_pm. if (year eq 1992) g5_92=g5*rnet_pm. if (year eq 1993) g5_93=g5*rnet_pm. if (year eq 1994) g5_94=g5*rnet_pm. if (year eq 1995) g5_95=g5*rnet_pm. if (year eq 1996) g5_96=g5*rnet_pm. if (year eq 1997) g5_97=g5*rnet_pm. if (year eq 1998) g5_98=g5*rnet_pm. if (year eq 1999) g5_99=g5*rnet_pm. if (year eq 2000) g5_00=g5*rnet_pm. if (year eq 2001) g5_01=g5*rnet_pm. if (year eq 1970) g6_70=g6*rnet_pm. if (year eq 1971) g6_71=g6*rnet_pm. if (year eq 1972) g6_72=g6*rnet_pm. if (year eq 1973) g6_73=g6*rnet_pm. if (year eq 1974) g6_74=g6*rnet_pm. if (year eq 1975) g6_75=g6*rnet_pm. if (year eq 1976) g6_76=g6*rnet_pm. if (year eq 1977) g6_77=g6*rnet_pm. if (year eq 1978) g6_78=g6*rnet_pm. if (year eq 1979) g6_79=g6*rnet_pm. if (year eq 1980) g6_80=g6*rnet_pm. if (year eq 1981) g6_81=g6*rnet_pm. if (year eq 1982) g6_82=g6*rnet_pm. if (year eq 1983) g6_83=g6*rnet_pm. if (year eq 1984) g6_84=g6*rnet_pm. if (year eq 1985) g6_85=g6*rnet_pm. if (year eq 1986) g6_86=g6*rnet_pm. if (year eq 1987) g6_87=g6*rnet_pm. if (year eq 1988) g6_88=g6*rnet_pm. if (year eq 1989) g6_89=g6*rnet_pm. if (year eq 1990) g6_90=g6*rnet_pm. if (year eq 1991) g6_91=g6*rnet_pm. if (year eq 1992) g6_92=g6*rnet_pm. if (year eq 1993) g6_93=g6*rnet_pm. if (year eq 1994) g6_94=g6*rnet_pm. if (year eq 1995) g6_95=g6*rnet_pm. if (year eq 1996) g6_96=g6*rnet_pm. if (year eq 1997) g6_97=g6*rnet_pm. if (year eq 1998) g6_98=g6*rnet_pm. if (year eq 1999) g6_99=g6*rnet_pm. if (year eq 2000) g6_00=g6*rnet_pm. if (year eq 2001) g6_01=g6*rnet_pm. if missing(g1_70) g1_70=0. if missing(g1_71) g1_71=0. if missing(g1_72) g1_72=0. if missing(g1_73) g1_73=0. if missing(g1_74) g1_74=0. if missing(g1_75) g1_75=0. if missing(g1_76) g1_76=0. if missing(g1_77) g1_77=0. if missing(g1_78) g1_78=0. if missing(g1_79) g1_79=0. if missing(g1_80) g1_80=0. if missing(g1_81) g1_81=0. if missing(g1_82) g1_82=0. if missing(g1_83) g1_83=0. if missing(g1_84) g1_84=0. if missing(g1_85) g1_85=0. if missing(g1_86) g1_86=0. if missing(g1_87) g1_87=0. if missing(g1_88) g1_88=0. if missing(g1_89) g1_89=0. if missing(g1_90) g1_90=0. if missing(g1_91) g1_91=0. if missing(g1_92) g1_92=0. if missing(g1_93) g1_93=0. if missing(g1_94) g1_94=0. if missing(g1_95) g1_95=0. if missing(g1_96) g1_96=0. if missing(g1_97) g1_97=0. if missing(g1_98) g1_98=0. if missing(g1_99) g1_99=0. if missing(g1_00) g1_00=0. if missing(g1_01) g1_01=0. if missing(g2_70) g2_70=0. if missing(g2_71) g2_71=0. if missing(g2_72) g2_72=0. if missing(g2_73) g2_73=0. if missing(g2_74) g2_74=0. if missing(g2_75) g2_75=0. if missing(g2_76) g2_76=0. if missing(g2_77) g2_77=0. if missing(g2_78) g2_78=0. if missing(g2_79) g2_79=0. if missing(g2_80) g2_80=0. if missing(g2_81) g2_81=0. 49 if missing(g2_82) g2_82=0. if missing(g2_83) g2_83=0. if missing(g2_84) g2_84=0. if missing(g2_85) g2_85=0. if missing(g2_86) g2_86=0. if missing(g2_87) g2_87=0. if missing(g2_88) g2_88=0. if missing(g2_89) g2_89=0. if missing(g2_90) g2_90=0. if missing(g2_91) g2_91=0. if missing(g2_92) g2_92=0. if missing(g2_93) g2_93=0. if missing(g2_94) g2_94=0. if missing(g2_95) g2_95=0. if missing(g2_96) g2_96=0. if missing(g2_97) g2_97=0. if missing(g2_98) g2_98=0. if missing(g2_99) g2_99=0. if missing(g2_00) g2_00=0. if missing(g2_01) g2_01=0. if missing(g3_70) g3_70=0. if missing(g3_71) g3_71=0. if missing(g3_72) g3_72=0. if missing(g3_73) g3_73=0. if missing(g3_74) g3_74=0. if missing(g3_75) g3_75=0. if missing(g3_76) g3_76=0. if missing(g3_77) g3_77=0. if missing(g3_78) g3_78=0. if missing(g3_79) g3_79=0. if missing(g3_80) g3_80=0. if missing(g3_81) g3_81=0. if missing(g3_82) g3_82=0. if missing(g3_83) g3_83=0. if missing(g3_84) g3_84=0. if missing(g3_85) g3_85=0. if missing(g3_86) g3_86=0. if missing(g3_87) g3_87=0. if missing(g3_88) g3_88=0. if missing(g3_89) g3_89=0. if missing(g3_90) g3_90=0. if missing(g3_91) g3_91=0. if missing(g3_92) g3_92=0. if missing(g3_93) g3_93=0. if missing(g3_94) g3_94=0. if missing(g3_95) g3_95=0. if missing(g3_96) g3_96=0. if missing(g3_97) g3_97=0. if missing(g3_98) g3_98=0. if missing(g3_99) g3_99=0. if missing(g3_00) g3_00=0. if missing(g3_01) g3_01=0. if missing(g4_70) g4_70=0. if missing(g4_71) g4_71=0. if missing(g4_72) g4_72=0. if missing(g4_73) g4_73=0. if missing(g4_74) g4_74=0. if missing(g4_75) g4_75=0. if missing(g4_76) g4_76=0. if missing(g4_77) g4_77=0. if missing(g4_78) g4_78=0. if missing(g4_79) g4_79=0. if missing(g4_80) g4_80=0. if missing(g4_81) g4_81=0. if missing(g4_82) g4_82=0. if missing(g4_83) g4_83=0. if missing(g4_84) g4_84=0. if missing(g4_85) g4_85=0. if missing(g4_86) g4_86=0. if missing(g4_87) g4_87=0. if missing(g4_88) g4_88=0. if missing(g4_89) g4_89=0. if missing(g4_90) g4_90=0. if missing(g4_91) g4_91=0. if missing(g4_92) g4_92=0. if missing(g4_93) g4_93=0. if missing(g4_94) g4_94=0. if missing(g4_95) g4_95=0. if missing(g4_96) g4_96=0. if missing(g4_97) g4_97=0. if missing(g4_98) g4_98=0. if missing(g4_99) g4_99=0. if missing(g4_00) g4_00=0. if missing(g4_01) g4_01=0. if missing(g5_70) g5_70=0. if missing(g5_71) g5_71=0. if missing(g5_72) g5_72=0. if missing(g5_73) g5_73=0. if missing(g5_74) g5_74=0. 50 if missing(g5_75) g5_75=0. if missing(g5_76) g5_76=0. if missing(g5_77) g5_77=0. if missing(g5_78) g5_78=0. if missing(g5_79) g5_79=0. if missing(g5_80) g5_80=0. if missing(g5_81) g5_81=0. if missing(g5_82) g5_82=0. if missing(g5_83) g5_83=0. if missing(g5_84) g5_84=0. if missing(g5_85) g5_85=0. if missing(g5_86) g5_86=0. if missing(g5_87) g5_87=0. if missing(g5_88) g5_88=0. if missing(g5_89) g5_89=0. if missing(g5_90) g5_90=0. if missing(g5_91) g5_91=0. if missing(g5_92) g5_92=0. if missing(g5_93) g5_93=0. if missing(g5_94) g5_94=0. if missing(g5_95) g5_95=0. if missing(g5_96) g5_96=0. if missing(g5_97) g5_97=0. if missing(g5_98) g5_98=0. if missing(g5_99) g5_99=0. if missing(g5_00) g5_00=0. if missing(g5_01) g5_01=0. if missing(g6_70) g6_70=0. if missing(g6_71) g6_71=0. if missing(g6_72) g6_72=0. if missing(g6_73) g6_73=0. if missing(g6_74) g6_74=0. if missing(g6_75) g6_75=0. if missing(g6_76) g6_76=0. if missing(g6_77) g6_77=0. if missing(g6_78) g6_78=0. if missing(g6_79) g6_79=0. if missing(g6_80) g6_80=0. if missing(g6_81) g6_81=0. if missing(g6_82) g6_82=0. if missing(g6_83) g6_83=0. if missing(g6_84) g6_84=0. if missing(g6_85) g6_85=0. if missing(g6_86) g6_86=0. if missing(g6_87) g6_87=0. if missing(g6_88) g6_88=0. if missing(g6_89) g6_89=0. if missing(g6_90) g6_90=0. if missing(g6_91) g6_91=0. if missing(g6_92) g6_92=0. if missing(g6_93) g6_93=0. if missing(g6_94) g6_94=0. if missing(g6_95) g6_95=0. if missing(g6_96) g6_96=0. if missing(g6_97) g6_97=0. if missing(g6_98) g6_98=0. if missing(g6_99) g6_99=0. if missing(g6_00) g6_00=0. if missing(g6_01) g6_01=0. aggregate outfile='h:\files\capital stock\flat.sav' /break=cso_ref /g1_70=sum(g1_70) /g1_71=sum(g1_71) /g1_72=sum(g1_72) /g1_73=sum(g1_73) /g1_74=sum(g1_74) /g1_75=sum(g1_75) /g1_76=sum(g1_76) /g1_77=sum(g1_77) /g1_78=sum(g1_78) /g1_79=sum(g1_79) /g1_80=sum(g1_80) /g1_81=sum(g1_81) /g1_82=sum(g1_82) /g1_83=sum(g1_83) /g1_84=sum(g1_84) /g1_85=sum(g1_85) /g1_86=sum(g1_86) /g1_87=sum(g1_87) /g1_88=sum(g1_88) /g1_89=sum(g1_89) /g1_90=sum(g1_90) /g1_91=sum(g1_91) /g1_92=sum(g1_92) /g1_93=sum(g1_93) /g1_94=sum(g1_94) /g1_95=sum(g1_95) /g1_96=sum(g1_96) 51 /g1_97=sum(g1_97) /g1_98=sum(g1_98) /g1_99=sum(g1_99) /g1_00=sum(g1_00) /g1_01=sum(g1_01) /g2_70=sum(g2_70) /g2_71=sum(g2_71) /g2_72=sum(g2_72) /g2_73=sum(g2_73) /g2_74=sum(g2_74) /g2_75=sum(g2_75) /g2_76=sum(g2_76) /g2_77=sum(g2_77) /g2_78=sum(g2_78) /g2_79=sum(g2_79) /g2_80=sum(g2_80) /g2_81=sum(g2_81) /g2_82=sum(g2_82) /g2_83=sum(g2_83) /g2_84=sum(g2_84) /g2_85=sum(g2_85) /g2_86=sum(g2_86) /g2_87=sum(g2_87) /g2_88=sum(g2_88) /g2_89=sum(g2_89) /g2_90=sum(g2_90) /g2_91=sum(g2_91) /g2_92=sum(g2_92) /g2_93=sum(g2_93) /g2_94=sum(g2_94) /g2_95=sum(g2_95) /g2_96=sum(g2_96) /g2_97=sum(g2_97) /g2_98=sum(g2_98) /g2_99=sum(g2_99) /g2_00=sum(g2_00) /g2_01=sum(g2_01) /g3_70=sum(g3_70) /g3_71=sum(g3_71) /g3_72=sum(g3_72) /g3_73=sum(g3_73) /g3_74=sum(g3_74) /g3_75=sum(g3_75) /g3_76=sum(g3_76) /g3_77=sum(g3_77) /g3_78=sum(g3_78) /g3_79=sum(g3_79) /g3_80=sum(g3_80) /g3_81=sum(g3_81) /g3_82=sum(g3_82) /g3_83=sum(g3_83) /g3_84=sum(g3_84) /g3_85=sum(g3_85) /g3_86=sum(g3_86) /g3_87=sum(g3_87) /g3_88=sum(g3_88) /g3_89=sum(g3_89) /g3_90=sum(g3_90) /g3_91=sum(g3_91) /g3_92=sum(g3_92) /g3_93=sum(g3_93) /g3_94=sum(g3_94) /g3_95=sum(g3_95) /g3_96=sum(g3_96) /g3_97=sum(g3_97) /g3_98=sum(g3_98) /g3_99=sum(g3_99) /g3_00=sum(g3_00) /g3_01=sum(g3_01) /g4_70=sum(g4_70) /g4_71=sum(g4_71) /g4_72=sum(g4_72) /g4_73=sum(g4_73) /g4_74=sum(g4_74) /g4_75=sum(g4_75) /g4_76=sum(g4_76) /g4_77=sum(g4_77) /g4_78=sum(g4_78) /g4_79=sum(g4_79) /g4_80=sum(g4_80) /g4_81=sum(g4_81) /g4_82=sum(g4_82) /g4_83=sum(g4_83) /g4_84=sum(g4_84) /g4_85=sum(g4_85) /g4_86=sum(g4_86) /g4_87=sum(g4_87) /g4_88=sum(g4_88) /g4_89=sum(g4_89) /g4_90=sum(g4_90) /g4_91=sum(g4_91) 52 /g4_92=sum(g4_92) /g4_93=sum(g4_93) /g4_94=sum(g4_94) /g4_95=sum(g4_95) /g4_96=sum(g4_96) /g4_97=sum(g4_97) /g4_98=sum(g4_98) /g4_99=sum(g4_99) /g4_00=sum(g4_00) /g4_01=sum(g4_01) /g5_70=sum(g5_70) /g5_71=sum(g5_71) /g5_72=sum(g5_72) /g5_73=sum(g5_73) /g5_74=sum(g5_74) /g5_75=sum(g5_75) /g5_76=sum(g5_76) /g5_77=sum(g5_77) /g5_78=sum(g5_78) /g5_79=sum(g5_79) /g5_80=sum(g5_80) /g5_81=sum(g5_81) /g5_82=sum(g5_82) /g5_83=sum(g5_83) /g5_84=sum(g5_84) /g5_85=sum(g5_85) /g5_86=sum(g5_86) /g5_87=sum(g5_87) /g5_88=sum(g5_88) /g5_89=sum(g5_89) /g5_90=sum(g5_90) /g5_91=sum(g5_91) /g5_92=sum(g5_92) /g5_93=sum(g5_93) /g5_94=sum(g5_94) /g5_95=sum(g5_95) /g5_96=sum(g5_96) /g5_97=sum(g5_97) /g5_98=sum(g5_98) /g5_99=sum(g5_99) /g5_00=sum(g5_00) /g5_01=sum(g5_01) /g6_70=sum(g6_70) /g6_71=sum(g6_71) /g6_72=sum(g6_72) /g6_73=sum(g6_73) /g6_74=sum(g6_74) /g6_75=sum(g6_75) /g6_76=sum(g6_76) /g6_77=sum(g6_77) /g6_78=sum(g6_78) /g6_79=sum(g6_79) /g6_80=sum(g6_80) /g6_81=sum(g6_81) /g6_82=sum(g6_82) /g6_83=sum(g6_83) /g6_84=sum(g6_84) /g6_85=sum(g6_85) /g6_86=sum(g6_86) /g6_87=sum(g6_87) /g6_88=sum(g6_88) /g6_89=sum(g6_89) /g6_90=sum(g6_90) /g6_91=sum(g6_91) /g6_92=sum(g6_92) /g6_93=sum(g6_93) /g6_94=sum(g6_94) /g6_95=sum(g6_95) /g6_96=sum(g6_96) /g6_97=sum(g6_97) /g6_98=sum(g6_98) /g6_99=sum(g6_99) /g6_00=sum(g6_00) /g6_01=sum(g6_01). get file='h:\files\capital stock\flat.sav'. compute gs6_01=g6_01+g6_00+g6_99+g6_98+g 6_97+g6_96+g6_95+g6_94+g6_93+g6_ 92+g6_91+g6_90+g6_89+g6_88+g6_87 +g6_86+g6_85+g6_84+ g6_83+g6_82+g6_81+g6_80+g6_79+g6 _78+g6_77+g6_76+g6_75+g6_74+g6_7 3+g6_72+g6_71+g6_70. compute gs6_00=gs6_01-g6_01. compute gs6_99=gs6_00-g6_00. compute gs6_98=gs6_99-g6_99. compute gs6_97=gs6_98-g6_98. compute gs6_96=gs6_97-g6_97. 53 compute gs6_95=gs6_96-g6_96. compute gs6_94=gs6_95-g6_95. compute gs6_93=gs6_94-g6_94. compute gs6_92=gs6_93-g6_93. compute gs6_91=gs6_92-g6_92. compute gs6_90=gs6_91-g6_91. compute gs6_89=gs6_90-g6_90. compute gs6_88=gs6_89-g6_89. compute gs6_87=gs6_88-g6_88. compute gs6_86=gs6_87-g6_87. compute gs6_85=gs6_86-g6_86. compute gs6_84=gs6_85-g6_85. compute gs6_83=gs6_84-g6_84. compute gs6_82=gs6_83-g6_83. compute gs6_81=gs6_82-g6_82. compute gs6_80=gs6_81-g6_81. compute gs6_79=gs6_80-g6_80. compute gs6_78=gs6_79-g6_79. compute gs6_77=gs6_78-g6_78. compute gs6_76=gs6_77-g6_77. compute gs6_75=gs6_76-g6_76. compute gs6_74=gs6_75-g6_75. compute gs6_73=gs6_74-g6_74. compute gs6_72=gs6_73-g6_73. compute gs6_71=gs6_72-g6_72. compute gs6_70=gs6_71-g6_71. compute gs5_01=g5_01+g5_00+g5_99+g5_98+g 5_97+g5_96+g5_95+g5_94+ g5_93+g5_92+g5_91+g5_90+g5_89+g5 _88+g5_87+g5_86+g5_85+g5_84+g5_8 3+g5_82+g5_81+ g5_80+g5_79+g5_78+g5_77+g5_76. compute gs5_00=gs5_01-g5_01+g5_75. compute gs5_99=gs5_00-g5_00+g5_74. compute gs5_98=gs5_99-g5_99+g5_73. compute gs5_97=gs5_98-g5_98+g5_72. compute gs5_96=gs5_97-g5_97+g5_71. compute gs5_95=gs5_96-g5_96+g5_70. compute gs5_94=gs5_95-g5_95. compute gs5_93=gs5_94-g5_94. compute gs5_92=gs5_93-g5_93. compute gs5_91=gs5_92-g5_92. compute gs5_90=gs5_91-g5_91. compute gs5_89=gs5_90-g5_90. compute gs5_88=gs5_89-g5_89. compute gs5_87=gs5_88-g5_88. compute gs5_86=gs5_87-g5_87. compute gs5_85=gs5_86-g5_86. compute gs5_84=gs5_85-g5_85. compute gs5_83=gs5_84-g5_84. compute gs5_82=gs5_83-g5_83. compute gs5_81=gs5_82-g5_82. compute gs5_80=gs5_81-g5_81. compute gs5_79=gs5_80-g5_80. compute gs5_78=gs5_79-g5_79. compute gs5_77=gs5_78-g5_78. compute gs5_76=gs5_77-g5_77. compute gs5_75=gs5_76-g5_76. compute gs5_74=gs5_75-g5_75. compute gs5_73=gs5_74-g5_74. compute gs5_72=gs5_73-g5_73. compute gs5_71=gs5_72-g5_72. compute gs5_70=gs5_71-g5_71. compute gs4_01=g4_01+g4_00+g4_99+g4_98+g 4_97+g4_96+g4_95+g4_94+ g4_93+g4_92+g4_91+g4_90+g4_89+g4 _88+g4_87+g4_86+g4_85+g4_84+g4_8 3. compute gs4_00=gs4_01-g4_01+g4_82. compute gs4_99=gs4_00-g4_00+g4_81. compute gs4_98=gs4_99-g4_99+g4_80. compute gs4_97=gs4_98-g4_98+g4_79. compute gs4_96=gs4_97-g4_97+g4_78. compute gs4_95=gs4_96-g4_96+g4_77. compute gs4_94=gs4_95-g4_95+g4_76. compute gs4_93=gs4_94-g4_94+g4_75. compute gs4_92=gs4_93-g4_93+g4_74. compute gs4_91=gs4_92-g4_92+g4_73. compute gs4_90=gs4_91-g4_91+g4_72. compute gs4_89=gs4_90-g4_90+g4_71. compute gs4_88=gs4_89-g4_89+g4_70. compute gs4_87=gs4_88-g4_88. compute gs4_86=gs4_87-g4_87. compute gs4_85=gs4_86-g4_86. compute gs4_84=gs4_85-g4_85. compute gs4_83=gs4_84-g4_84. 54 compute gs4_82=gs4_83-g4_83. compute gs4_81=gs4_82-g4_82. compute gs4_80=gs4_81-g4_81. compute gs4_79=gs4_80-g4_80. compute gs4_78=gs4_79-g4_79. compute gs4_77=gs4_78-g4_78. compute gs4_76=gs4_77-g4_77. compute gs4_75=gs4_76-g4_76. compute gs4_74=gs4_75-g4_75. compute gs4_73=gs4_74-g4_74. compute gs4_72=gs4_73-g4_73. compute gs4_71=gs4_72-g4_72. compute gs4_70=gs4_71-g4_71. compute gs3_01=g3_01+g3_00+g3_99+g3_98+g 3_97+g3_96+g3_95+g3_94+ g3_93+g3_92+g3_91+g3_90+g3_89+g3 _88. compute gs3_00=gs3_01-g3_01+g3_87. compute gs3_99=gs3_00-g3_00+g3_86. compute gs3_98=gs3_99-g3_99+g3_85. compute gs3_97=gs3_98-g3_98+g3_84. compute gs3_96=gs3_97-g3_97+g3_83. compute gs3_95=gs3_96-g3_96+g3_82. compute gs3_94=gs3_95-g3_95+g3_81. compute gs3_93=gs3_94-g3_94+g3_80. compute gs3_92=gs3_93-g3_93+g3_79. compute gs3_91=gs3_92-g3_92+g3_78. compute gs3_90=gs3_91-g3_91+g3_77. compute gs3_89=gs3_90-g3_90+g3_76. compute gs3_88=gs3_89-g3_89+g3_75. compute gs3_87=gs3_88-g3_88+g3_74. compute gs3_86=gs3_87-g3_87+g3_73. compute gs3_85=gs3_86-g3_86+g3_72. compute gs3_84=gs3_85-g3_85+g3_71. compute gs3_83=gs3_84-g3_84+g3_70. compute gs3_82=gs3_83-g3_83. compute gs3_81=gs3_82-g3_82. compute gs3_80=gs3_81-g3_81. compute gs3_79=gs3_80-g3_80. compute gs3_78=gs3_79-g3_79. compute gs3_77=gs3_78-g3_78. compute gs3_76=gs3_77-g3_77. compute gs3_75=gs3_76-g3_76. compute gs3_74=gs3_75-g3_75. compute gs3_73=gs3_74-g3_74. compute gs3_72=gs3_73-g3_73. compute gs3_71=gs3_72-g3_72. compute gs3_70=gs3_71-g3_71. compute gs2_01=g2_01+g2_00+g2_99+g2_98+g 2_97+g2_96+g2_95+g2_94+ g2_93+g2_92+g2_91+g2_90. compute gs2_00=gs2_01-g2_01+g2_89. compute gs2_99=gs2_00-g2_00+g2_88. compute gs2_98=gs2_99-g2_99+g2_87. compute gs2_97=gs2_98-g2_98+g2_86. compute gs2_96=gs2_97-g2_97+g2_85. compute gs2_95=gs2_96-g2_96+g2_84. compute gs2_94=gs2_95-g2_95+g2_83. compute gs2_93=gs2_94-g2_94+g2_82. compute gs2_92=gs2_93-g2_93+g2_81. compute gs2_91=gs2_92-g2_92+g2_80. compute gs2_90=gs2_91-g2_91+g2_79. compute gs2_89=gs2_90-g2_90+g2_78. compute gs2_88=gs2_89-g2_89+g2_77. compute gs2_87=gs2_88-g2_88+g2_76. compute gs2_86=gs2_87-g2_87+g2_75. compute gs2_85=gs2_86-g2_86+g2_74. compute gs2_84=gs2_85-g2_85+g2_73. compute gs2_83=gs2_84-g2_84+g2_72. compute gs2_82=gs2_83-g2_83+g2_71. compute gs2_81=gs2_82-g2_82+g2_70. compute gs2_80=gs2_81-g2_81. compute gs2_79=gs2_80-g2_80. compute gs2_78=gs2_79-g2_79. compute gs2_77=gs2_78-g2_78. compute gs2_76=gs2_77-g2_77. compute gs2_75=gs2_76-g2_76. compute gs2_74=gs2_75-g2_75. compute gs2_73=gs2_74-g2_74. compute gs2_72=gs2_73-g2_73. compute gs2_71=gs2_72-g2_72. compute gs2_70=gs2_71-g2_71. compute gs1_01=g1_01+g1_00+g1_99+g1_98+g 1_97. compute gs1_00=gs1_01-g1_01+g1_96. 55 compute gs1_99=gs1_00-g1_00+g1_95. compute gs1_98=gs1_99-g1_99+g1_94. compute gs1_97=gs1_98-g1_98+g1_93. compute gs1_96=gs1_97-g1_97+g1_92. compute gs1_95=gs1_96-g1_96+g1_91. compute gs1_94=gs1_95-g1_95+g1_90. compute gs1_93=gs1_94-g1_94+g1_89. compute gs1_92=gs1_93-g1_93+g1_88. compute gs1_91=gs1_92-g1_92+g1_87. compute gs1_90=gs1_91-g1_91+g1_86. compute gs1_89=gs1_90-g1_90+g1_85. compute gs1_88=gs1_89-g1_89+g1_84. compute gs1_87=gs1_88-g1_88+g1_83. compute gs1_86=gs1_87-g1_87+g1_82. compute gs1_85=gs1_86-g1_86+g1_81. compute gs1_84=gs1_85-g1_85+g1_80. compute gs1_83=gs1_84-g1_84+g1_79. compute gs1_82=gs1_83-g1_83+g1_78. compute gs1_81=gs1_82-g1_82+g1_77. compute gs1_80=gs1_81-g1_81+g1_76. compute gs1_79=gs1_80-g1_80+g1_75. compute gs1_78=gs1_79-g1_79+g1_74. compute gs1_77=gs1_78-g1_78+g1_73. compute gs1_76=gs1_77-g1_77+g1_72. compute gs1_75=gs1_76-g1_76+g1_71. compute gs1_74=gs1_75-g1_75+g1_70. compute gs1_73=gs1_74-g1_74. compute gs1_72=gs1_73-g1_73. compute gs1_71=gs1_72-g1_72. compute gs1_70=gs1_71-g1_71. compute ns6_01=.973*g6_01+.9459*g6_00+.918 9*g6_99+.8919*g6_98+.8649*g6_97+.8 378*g6_96+.8108*g6_95+ .7838*g6_94+.7567*g6_93+.7297*g6_9 2+.7027*g6_91+.6757*g6_90+.6486*g6 _89+ .6216*g6_88+.5946*g6_87+.5676*g6_8 6+.5405*g6_85+.5135*g6_84+.4865*g6 _83+.4595*g6_82+.4324*g6_81+.4054* g6_80+ .3783*g6_79+.3514*g6_78+.3244*g6_7 7+.2973*g6_76+.2703*g6_75+.2433*g6 _74+.2163*g6_73+.1893*g6_72+.1623* g6_71+.1352*g6_70. compute ns6_00=.973*g6_00+.9459*g6_99+.918 9*g6_98+.8919*g6_97+.8649*g6_96+.8 378*g6_95+.8108*g6_94+ .7838*g6_93+.7567*g6_92+.7297*g6_9 1+.7027*g6_90+.6757*g6_89+.6486*g6 _88+ .6216*g6_87+.5946*g6_86+.5676*g6_8 5+.5405*g6_84+.5135*g6_83+.4865*g6 _82+.4595*g6_81+.4324*g6_80+.4054* g6_79+ .3783*g6_78+.3514*g6_77+.3244*g6_7 6+.2973*g6_75+.2703*g6_74+.2433*g6 _73+.2163*g6_72+.1893*g6_71+.1623* g6_70. compute ns6_99=.973*g6_99+.9459*g6_98+.918 9*g6_97+.8919*g6_96+.8649*g6_95+.8 378*g6_94+.8108*g6_93+ .7838*g6_92+.7567*g6_91+.7297*g6_9 0+.7027*g6_89+.6757*g6_88+.6486*g6 _87+ .6216*g6_86+.5946*g6_85+.5676*g6_8 4+.5405*g6_83+.5135*g6_82+.4865*g6 _81+.4595*g6_80+.4324*g6_79+.4054* g6_78+ .3783*g6_77+.3514*g6_76+.3244*g6_7 5+.2973*g6_74+.2703*g6_73+.2433*g6 _72+.2163*g6_71+.1893*g6_70. compute ns6_98=.973*g6_98+.9459*g6_97+.918 9*g6_96+.8919*g6_95+.8649*g6_94+.8 378*g6_93+.8108*g6_92+ .7838*g6_91+.7567*g6_90+.7297*g6_8 9+.7027*g6_88+.6757*g6_87+.6486*g6 _86+ 56 .6216*g6_85+.5946*g6_84+.5676*g6_8 3+.5405*g6_82+.5135*g6_81+.4865*g6 _80+.4595*g6_79+.4324*g6_78+.4054* g6_77+ .3783*g6_76+.3514*g6_75+.3244*g6_7 4+.2973*g6_73+.2703*g6_72+.2433*g6 _71+.2163*g6_70. compute ns6_97=.973*g6_97+.9459*g6_96+.918 9*g6_95+.8919*g6_94+.8649*g6_93+.8 378*g6_92+.8108*g6_91+ .7838*g6_90+.7567*g6_89+.7297*g6_8 8+.7027*g6_87+.6757*g6_86+.6486*g6 _85+ .6216*g6_84+.5946*g6_83+.5676*g6_8 2+.5405*g6_81+.5135*g6_80+.4865*g6 _79+.4595*g6_78+.4324*g6_77+.4054* g6_76+ .3783*g6_75+.3514*g6_74+.3244*g6_7 3+.2973*g6_72+.2703*g6_71+.2433*g6 _70. compute ns6_96=.973*g6_96+.9459*g6_95+.918 9*g6_94+.8919*g6_93+.8649*g6_92+.8 378*g6_91+.8108*g6_90+ .7838*g6_89+.7567*g6_88+.7297*g6_8 7+.7027*g6_86+.6757*g6_85+.6486*g6 _84+ .6216*g6_83+.5946*g6_82+.5676*g6_8 1+.5405*g6_80+.5135*g6_79+.4865*g6 _78+.4595*g6_77+.4324*g6_76+.4054* g6_75+ .3783*g6_74+.3514*g6_73+.3244*g6_7 2+.2973*g6_71+.2703*g6_70. compute ns6_95=.973*g6_95+.9459*g6_94+.918 9*g6_93+.8919*g6_92+.8649*g6_91+.8 378*g6_90+.8108*g6_89+ .7838*g6_88+.7567*g6_87+.7297*g6_8 6+.7027*g6_85+.6757*g6_84+.6486*g6 _83+ .6216*g6_82+.5946*g6_81+.5676*g6_8 0+.5405*g6_79+.5135*g6_78+.4865*g6 _77+.4595*g6_76+.4324*g6_75+.4054* g6_74+ .3783*g6_73+.3514*g6_72+.3244*g6_7 1+.2973*g6_70. compute ns6_94=.973*g6_94+.9459*g6_93+.918 9*g6_92+.8919*g6_91+.8649*g6_90+.8 378*g6_89+.8108*g6_88+ .7838*g6_87+.7567*g6_86+.7297*g6_8 5+.7027*g6_84+.6757*g6_83+.6486*g6 _82+ .6216*g6_81+.5946*g6_80+.5676*g6_7 9+.5405*g6_78+.5135*g6_77+.4865*g6 _76+.4595*g6_75+.4324*g6_74+.4054* g6_73+ .3783*g6_72+.3514*g6_71+.3244*g6_7 0. compute ns6_93=.973*g6_93+.9459*g6_92+.918 9*g6_91+.8919*g6_90+.8649*g6_89+.8 378*g6_88+.8108*g6_87+ .7838*g6_86+.7567*g6_85+.7297*g6_8 4+.7027*g6_83+.6757*g6_82+.6486*g6 _81+ .6216*g6_80+.5946*g6_79+.5676*g6_7 8+.5405*g6_77+.5135*g6_76+.4865*g6 _75+.4595*g6_74+.4324*g6_73+.4054* g6_72+ .3783*g6_71+.3514*g6_70. compute ns6_92=.973*g6_92+.9459*g6_91+.918 9*g6_90+.8919*g6_89+.8649*g6_88+.8 378*g6_87+.8108*g6_86+ 57 .7838*g6_85+.7567*g6_84+.7297*g6_8 3+.7027*g6_82+.6757*g6_81+.6486*g6 _80+ .6216*g6_79+.5946*g6_78+.5676*g6_7 7+.5405*g6_76+.5135*g6_75+.4865*g6 _74+.4595*g6_73+.4324*g6_72+.4054* g6_71+ .3783*g6_70. compute ns6_91=.973*g6_91+.9459*g6_90+.918 9*g6_89+.8919*g6_88+.8649*g6_87+.8 378*g6_86+.8108*g6_85+ .7838*g6_84+.7567*g6_83+.7297*g6_8 2+.7027*g6_81+.6757*g6_80+.6486*g6 _79+ .6216*g6_78+.5946*g6_77+.5676*g6_7 6+.5405*g6_75+.5135*g6_74+.4865*g6 _73+.4595*g6_72+.4324*g6_71+.4054* g6_70. compute ns6_90=.973*g6_90+.9459*g6_89+.918 9*g6_88+.8919*g6_87+.8649*g6_86+.8 378*g6_85+.8108*g6_84+ .7838*g6_83+.7567*g6_82+.7297*g6_8 1+.7027*g6_80+.6757*g6_79+.6486*g6 _78+ .6216*g6_77+.5946*g6_76+.5676*g6_7 5+.5405*g6_74+.5135*g6_73+.4865*g6 _72+.4595*g6_71+.4324*g6_70. compute ns6_89=.973*g6_89+.9459*g6_88+.918 9*g6_87+.8919*g6_86+.8649*g6_85+.8 378*g6_84+.8108*g6_83+ .7838*g6_82+.7567*g6_81+.7297*g6_8 0+.7027*g6_79+.6757*g6_78+.6486*g6 _77+ 4+.5405*g6_73+.5135*g6_72+.4865*g6 _71+.4595*g6_70. compute ns6_88=.973*g6_88+.9459*g6_87+.918 9*g6_86+.8919*g6_85+.8649*g6_84+.8 378*g6_83+.8108*g6_82+ .7838*g6_81+.7567*g6_80+.7297*g6_7 9+.7027*g6_78+.6757*g6_77+.6486*g6 _78+ .6216*g6_75+.5946*g6_74+.5676*g6_7 3+.5405*g6_72+.5135*g6_71+.4865*g6 _70. compute ns6_87=.973*g6_87+.9459*g6_86+.918 9*g6_85+.8919*g6_84+.8649*g6_83+.8 378*g6_82+.8108*g6_81+ .7838*g6_80+.7567*g6_79+.7297*g6_7 8+.7027*g6_77+.6757*g6_76+.6486*g6 _75+ .6216*g6_74+.5946*g6_73+.5676*g6_7 2+.5405*g6_71+.5135*g6_70. compute ns6_86=.973*g6_86+.9459*g6_85+.918 9*g6_84+.8919*g6_83+.8649*g6_82+.8 378*g6_81+.8108*g6_80+ .7838*g6_79+.7567*g6_78+.7297*g6_7 7+.7027*g6_76+.6757*g6_75+.6486*g6 _74+ .6216*g6_73+.5946*g6_72+.5676*g6_7 1+.5405*g6_70. compute ns6_85=.973*g6_85+.9459*g6_84+.918 9*g6_83+.8919*g6_82+.8649*g6_81+.8 378*g6_80+.8108*g6_79+ .7838*g6_78+.7567*g6_77+.7297*g6_7 6+.7027*g6_75+.6757*g6_74+.6486*g6 _73+ .6216*g6_76+.5946*g6_75+.5676*g6_7 58 .6216*g6_72+.5946*g6_71+.5676*g6_7 0. compute ns6_84=.973*g6_84+.9459*g6_83+.918 9*g6_82+.8919*g6_81+.8649*g6_80+.8 378*g6_79+.8108*g6_78+ .7838*g6_77+.7567*g6_76+.7297*g6_7 5+.7027*g6_74+.6757*g6_73+.6486*g6 _72+ .6216*g6_71+.5946*g6_70. compute ns6_83=.973*g6_83+.9459*g6_82+.918 9*g6_81+.8919*g6_80+.8649*g6_79+.8 378*g6_78+.8108*g6_77+ .7838*g6_76+.7567*g6_75+.7297*g6_7 4+.7027*g6_73+.6757*g6_72+.6486*g6 _71+ .6216*g6_70. compute ns6_82=.973*g6_82+.9459*g6_81+.918 9*g6_80+.8919*g6_79+.8649*g6_78+.8 378*g6_77+.8108*g6_76+ .7838*g6_75+.7567*g6_74+.7297*g6_7 3+.7027*g6_72+.6757*g6_71+.6486*g6 _70. compute ns6_81=.973*g6_81+.9459*g6_80+.918 9*g6_79+.8919*g6_78+.8649*g6_77+.8 378*g6_76+.8108*g6_75+ .7838*g6_74+.7567*g6_73+.7297*g6_7 2+.7027*g6_71+.6757*g6_70. compute ns6_80=.973*g6_80+.9459*g6_79+.918 9*g6_78+.8919*g6_77+.8649*g6_76+.8 378*g6_75+.8108*g6_74+ .7838*g6_73+.7567*g6_72+.7297*g6_7 1+.7027*g6_70. compute ns6_79=.973*g6_79+.9459*g6_78+.918 9*g6_77+.8919*g6_76+.8649*g6_75+.8 378*g6_74+.8108*g6_73+ .7838*g6_72+.7567*g6_71+.7297*g6_7 0. compute ns6_78=.973*g6_78+.9459*g6_77+.918 9*g6_76+.8919*g6_75+.8649*g6_74+.8 378*g6_73+.8108*g6_72+ .7838*g6_71+.7567*g6_70. compute ns6_77=.973*g6_77+.9459*g6_76+.918 9*g6_75+.8919*g6_74+.8649*g6_73+.8 378*g6_72+.8108*g6_71+ .7838*g6_70. compute ns6_76=.973*g6_76+.9459*g6_75+.918 9*g6_74+.8919*g6_73+.8649*g6_72+.8 378*g6_71+.8108*g6_70. compute ns6_75=.973*g6_75+.9459*g6_74+.918 9*g6_73+.8919*g6_72+.8649*g6_71+.8 378*g6_70. compute ns6_74=.973*g6_74+.9459*g6_73+.918 9*g6_72+.8919*g6_71+.8649*g6_70. compute ns6_73=.973*g6_73+.9459*g6_72+.918 9*g6_71+.8919*g6_70. compute ns6_72=.973*g6_72+.9459*g6_71+.918 9*g6_70. compute ns6_71=.973*g6_71+.9459*g6_70. compute ns6_70=.973*g6_70. compute ns5_01=.9615*g5_01+.9231*g5_00+.88 46*g5_99+.8462*g5_98+.8077*g5_97+. 7692*g5_96+.7308*g5_95+ .6923*g5_94+.6538*g5_93+.6154*g5_9 2+.5769*g5_91+.5385*g5_90+.5*g5_89 + .4615*g5_88+.4231*g5_87+.3846*g5_8 59 6+.3462*g5_85+.3077*g5_84+.2692*g5 _83+.2308*g5_82+.1923*g5_81+.1538* g5_80+ .1154*g5_79+.0769*g5_78+.03848*g5_ 77. compute ns5_00=.9615*g5_00+.9231*g5_99+.88 46*g5_98+.8462*g5_97+.8077*g5_96+. 7692*g5_95+.7308*g5_94+ .6923*g5_93+.6538*g5_92+.6154*g5_9 1+.5769*g5_90+.5385*g5_89+.5*g5_88 + .4615*g5_87+.4231*g5_86+.3846*g5_8 5+.3462*g5_84+.3077*g5_83+.2692*g5 _82+.2308*g5_81+.1923*g5_80+.1538* g5_79+ .1154*g5_78+.0769*g5_77+.03848*g5_ 76. compute ns5_99=.9615*g5_99+.9231*g5_98+.88 46*g5_97+.8462*g5_96+.8077*g5_95+. 7692*g5_94+.7308*g5_93+ .6923*g5_92+.6538*g5_91+.6154*g5_9 0+.5769*g5_89+.5385*g5_88+.5*g5_87 + .4615*g5_86+.4231*g5_85+.3846*g5_8 4+.3462*g5_83+.3077*g5_82+.2692*g5 _81+.2308*g5_80+.1923*g5_79+.1538* g5_78+ .1154*g5_77+.0769*g5_76+.03848*g5_ 75. compute ns5_98=.9615*g5_98+.9231*g5_97+.88 46*g5_96+.8462*g5_95+.8077*g5_94+. 7692*g5_93+.7308*g5_92+ .6923*g5_91+.6538*g5_90+.6154*g5_8 9+.5769*g5_88+.5385*g5_87+.5*g5_86 + .4615*g5_85+.4231*g5_84+.3846*g5_8 3+.3462*g5_82+.3077*g5_81+.2692*g5 _80+.2308*g5_79+.1923*g5_78+.1538* g5_77+ .1154*g5_76+.0769*g5_75+.03848*g5_ 74. compute ns5_97=.9615*g5_97+.9231*g5_96+.88 46*g5_95+.8462*g5_94+.8077*g5_93+. 7692*g5_92+.7308*g5_91+ .6923*g5_90+.6538*g5_89+.6154*g5_8 8+.5769*g5_87+.5385*g5_86+.5*g5_85 + .4615*g5_84+.4231*g5_83+.3846*g5_8 2+.3462*g5_81+.3077*g5_80+.2692*g5 _79+.2308*g5_78+.1923*g5_77+.1538* g5_76+ .1154*g5_75+.0769*g5_74+.03848*g5_ 73. compute ns5_96=.9615*g5_96+.9231*g5_95+.88 46*g5_94+.8462*g5_93+.8077*g5_92+. 7692*g5_91+.7308*g5_90+ .6923*g5_89+.6538*g5_88+.6154*g5_8 7+.5769*g5_86+.5385*g5_85+.5*g5_84 + .4615*g5_83+.4231*g5_82+.3846*g5_8 1+.3462*g5_80+.3077*g5_79+.2692*g5 _78+.2308*g5_77+.1923*g5_76+.1538* g5_75+ .1154*g5_74+.0769*g5_73+.03848*g5_ 72. compute ns5_95=.9615*g5_95+.9231*g5_94+.88 46*g5_93+.8462*g5_92+.8077*g5_91+. 7692*g5_90+.7308*g5_89+ .6923*g5_88+.6538*g5_87+.6154*g5_8 60 6+.5769*g5_85+.5385*g5_84+.5*g5_83 + 3+.5769*g5_82+.5385*g5_81+.5*g5_80 + .4615*g5_82+.4231*g5_81+.3846*g5_8 0+.3462*g5_79+.3077*g5_78+.2692*g5 _77+.2308*g5_76+.1923*g5_75+.1538* g5_74+ .4615*g5_79+.4231*g5_78+.3846*g5_7 7+.3462*g5_76+.3077*g5_75+.2692*g5 _74+.2308*g5_73+.1923*g5_72+.1538* g5_71+ .1154*g5_70. compute ns5_91=.9615*g5_91+.9231*g5_90+.88 46*g5_89+.8462*g5_88+.8077*g5_87+. 7692*g5_86+.7308*g5_85+ .1154*g5_73+.0769*g5_72+.03848*g5_ 71. compute ns5_94=.9615*g5_94+.9231*g5_93+.88 46*g5_92+.8462*g5_91+.8077*g5_90+. 7692*g5_89+.7308*g5_88+ .6923*g5_87+.6538*g5_86+.6154*g5_8 5+.5769*g5_84+.5385*g5_83+.5*g5_82 + .4615*g5_81+.4231*g5_80+.3846*g5_7 9+.3462*g5_78+.3077*g5_77+.2692*g5 _76+.2308*g5_75+.1923*g5_74+.1538* g5_73+ .1154*g5_72+.0769*g5_71+.03848*g5_ 70. compute ns5_93=.9615*g5_93+.9231*g5_92+.88 46*g5_91+.8462*g5_90+.8077*g5_89+. 7692*g5_88+.7308*g5_87+ .6923*g5_86+.6538*g5_85+.6154*g5_8 4+.5769*g5_83+.5385*g5_82+.5*g5_81 + .4615*g5_80+.4231*g5_79+.3846*g5_7 8+.3462*g5_77+.3077*g5_76+.2692*g5 _75+.2308*g5_74+.1923*g5_73+.1538* g5_72+ .1154*g5_71+.0769*g5_70. compute ns5_92=.9615*g5_92+.9231*g5_91+.88 46*g5_90+.8462*g5_89+.8077*g5_88+. 7692*g5_87+.7308*g5_86+ .6923*g5_84+.6538*g5_83+.6154*g5_8 2+.5769*g5_81+.5385*g5_80+.5*g5_79 + .4615*g5_78+.4231*g5_77+.3846*g5_7 6+.3462*g5_75+.3077*g5_74+.2692*g5 _73+.2308*g5_72+.1923*g5_71+.1538* g5_70. compute ns5_90=.9615*g5_90+.9231*g5_89+.88 46*g5_88+.8462*g5_87+.8077*g5_86+. 7692*g5_85+.7308*g5_84+ .6923*g5_83+.6538*g5_82+.6154*g5_8 1+.5769*g5_80+.5385*g5_79+.5*g5_78 + .4615*g5_77+.4231*g5_76+.3846*g5_7 5+.3462*g5_74+.3077*g5_73+.2692*g5 _72+.2308*g5_71+.1923*g5_70. compute ns5_89=.9615*g5_89+.9231*g5_88+.88 46*g5_87+.8462*g5_86+.8077*g5_85+. 7692*g5_84+.7308*g5_83+ .6923*g5_82+.6538*g5_81+.6154*g5_8 0+.5769*g5_79+.5385*g5_78+.5*g5_77 + .4615*g5_76+.4231*g5_75+.3846*g5_7 4+.3462*g5_73+.3077*g5_72+.2692*g5 _71+.2308*g5_70. .6923*g5_85+.6538*g5_84+.6154*g5_8 61 compute ns5_88=.9615*g5_88+.9231*g5_87+.88 46*g5_86+.8462*g5_85+.8077*g5_84+. 7692*g5_83+.7308*g5_82+ compute ns5_84=.9615*g5_84+.9231*g5_83+.88 46*g5_82+.8462*g5_81+.8077*g5_80+. 7692*g5_79+.7308*g5_78+ .6923*g5_81+.6538*g5_80+.6154*g5_7 9+.5769*g5_78+.5385*g5_77+.5*g5_78 + .6923*g5_77+.6538*g5_76+.6154*g5_7 5+.5769*g5_74+.5385*g5_73+.5*g5_72 + .4615*g5_71+.4231*g5_70. compute ns5_83=.9615*g5_83+.9231*g5_82+.88 46*g5_81+.8462*g5_80+.8077*g5_79+. 7692*g5_78+.7308*g5_77+ .4615*g5_75+.4231*g5_74+.3846*g5_7 3+.3462*g5_72+.3077*g5_71+.2692*g5 _70. compute ns5_87=.9615*g5_87+.9231*g5_86+.88 46*g5_85+.8462*g5_84+.8077*g5_83+. 7692*g5_82+.7308*g5_81+ .6923*g5_80+.6538*g5_79+.6154*g5_7 8+.5769*g5_77+.5385*g5_76+.5*g5_75 + .4615*g5_74+.4231*g5_73+.3846*g5_7 2+.3462*g5_71+.3077*g5_70. compute ns5_86=.9615*g5_86+.9231*g5_85+.88 46*g5_84+.8462*g5_83+.8077*g5_82+. 7692*g5_81+.7308*g5_80+ .6923*g5_79+.6538*g5_78+.6154*g5_7 7+.5769*g5_76+.5385*g5_75+.5*g5_74 + .4615*g5_73+.4231*g5_72+.3846*g5_7 1+.3462*g5_70. compute ns5_85=.9615*g5_85+.9231*g5_84+.88 46*g5_83+.8462*g5_82+.8077*g5_81+. 7692*g5_80+.7308*g5_79+ .6923*g5_78+.6538*g5_77+.6154*g5_7 6+.5769*g5_75+.5385*g5_74+.5*g5_73 + .4615*g5_72+.4231*g5_71+.3846*g5_7 0. .6923*g5_76+.6538*g5_75+.6154*g5_7 4+.5769*g5_73+.5385*g5_72+.5*g5_71 + .4615*g5_70. compute ns5_82=.9615*g5_82+.9231*g5_81+.88 46*g5_80+.8462*g5_79+.8077*g5_78+. 7692*g5_77+.7308*g5_76+ .6923*g5_75+.6538*g5_74+.6154*g5_7 3+.5769*g5_72+.5385*g5_71+.5*g5_70 . compute ns5_81=.9615*g5_81+.9231*g5_80+.88 46*g5_79+.8462*g5_78+.8077*g5_77+. 7692*g5_76+.7308*g5_75+ .6923*g5_74+.6538*g5_73+.6154*g5_7 2+.5769*g5_71+.5385*g5_70. compute ns5_80=.9615*g5_80+.9231*g5_79+.88 46*g5_78+.8462*g5_77+.8077*g5_76+. 7692*g5_75+.7308*g5_74+ .6923*g5_73+.6538*g5_72+.6154*g5_7 1+.5769*g5_70. compute ns5_79=.9615*g5_79+.9231*g5_78+.88 46*g5_77+.8462*g5_76+.8077*g5_75+. 7692*g5_74+.7308*g5_73+ 62 .6923*g5_72+.6538*g5_71+.6154*g5_7 0. compute ns5_78=.9615*g5_78+.9231*g5_77+.88 46*g5_76+.8462*g5_75+.8077*g5_74+. 7692*g5_73+.7308*g5_72+ .6923*g5_71+.6538*g5_70. compute ns5_77=.9615*g5_77+.9231*g5_76+.88 46*g5_75+.8462*g5_74+.8077*g5_73+. 7692*g5_72+.7308*g5_71+ .6923*g5_70. compute ns5_76=.9615*g5_76+.9231*g5_75+.88 46*g5_74+.8462*g5_73+.8077*g5_72+. 7692*g5_71+.7308*g5_70. compute ns5_75=.9615*g5_75+.9231*g5_74+.88 46*g5_73+.8462*g5_72+.8077*g5_71+. 7692*g5_70. compute ns5_74=.9615*g5_74+.9231*g5_73+.88 46*g5_72+.8462*g5_71+.8077*g5_70. compute ns5_73=.9615*g5_73+.9231*g5_72+.88 46*g5_71+.8462*g5_70. compute ns5_72=.9615*g5_72+.9231*g5_71+.88 46*g5_70. compute ns5_71=.9615*g5_71+.9231*g5_70. compute ns5_70=.9615*g5_70. compute ns4_01=.9474*g4_01+.8947*g4_00+.84 21*g4_99+.7895*g4_98+.7368*g4_97+. 6842*g4_96+.6316*g4_95+ .5789*g4_94+.5263*g4_93+.4737*g4_9 2+.421*g4_91+.3684*g4_90+.3158*g4_ 89+.2632*g4_88+.2105*g4_87+ .1579*g4_86+.1053*g4_85+.0526*g4_8 4. compute ns4_00=.9474*g4_00+.8947*g4_99+.84 21*g4_98+.7895*g4_97+.7368*g4_96+. 6842*g4_95+.6316*g4_94+ .5789*g4_93+.5263*g4_92+.4737*g4_9 1+.421*g4_90+.3684*g4_89+.3158*g4_ 88+.2632*g4_87+.2105*g4_86+ .1579*g4_85+.1053*g4_84+.0526*g4_8 3. compute ns4_99=.9474*g4_99+.8947*g4_98+.84 21*g4_97+.7895*g4_96+.7368*g4_95+. 6842*g4_94+.6316*g4_93+ .5789*g4_92+.5263*g4_91+.4737*g4_9 0+.421*g4_89+.3684*g4_88+.3158*g4_ 87+.2632*g4_86+.2105*g4_85+ .1579*g4_84+.1053*g4_83+.0526*g4_8 2. compute ns4_98=.9474*g4_98+.8947*g4_97+.84 21*g4_96+.7895*g4_95+.7368*g4_94+. 6842*g4_93+.6316*g4_92+ .5789*g4_91+.5263*g4_90+.4737*g4_8 9+.421*g4_88+.3684*g4_87+.3158*g4_ 86+.2632*g4_85+.2105*g4_84+ .1579*g4_83+.1053*g4_82+.0526*g4_8 1. compute ns4_97=.9474*g4_97+.8947*g4_96+.84 21*g4_95+.7895*g4_94+.7368*g4_93+. 6842*g4_92+.6316*g4_91+ .5789*g4_90+.5263*g4_89+.4737*g4_8 8+.421*g4_87+.3684*g4_86+.3158*g4_ 85+.2632*g4_84+.2105*g4_83+ .1579*g4_82+.1053*g4_81+.0526*g4_8 0. compute ns4_96=.9474*g4_96+.8947*g4_95+.84 63 21*g4_94+.7895*g4_93+.7368*g4_92+. 6842*g4_91+.6316*g4_90+ .5789*g4_89+.5263*g4_88+.4737*g4_8 7+.421*g4_86+.3684*g4_85+.3158*g4_ 84+.2632*g4_83+.2105*g4_82+ .1579*g4_81+.1053*g4_80+.0526*g4_7 9. compute ns4_95=.9474*g4_95+.8947*g4_94+.84 21*g4_93+.7895*g4_92+.7368*g4_91+. 6842*g4_90+.6316*g4_89+ .5789*g4_88+.5263*g4_87+.4737*g4_8 6+.421*g4_85+.3684*g4_84+.3158*g4_ 83+.2632*g4_82+.2105*g4_81+ .1579*g4_80+.1053*g4_79+.0526*g4_7 8+0*g4_77. compute ns4_94=.9474*g4_94+.8947*g4_93+.84 21*g4_92+.7895*g4_91+.7368*g4_90+. 6842*g4_89+.6316*g4_88+ .5789*g4_87+.5263*g4_86+.4737*g4_8 5+.421*g4_84+.3684*g4_83+.3158*g4_ 82+.2632*g4_81+.2105*g4_80+ .1579*g4_79+.1053*g4_78+.0526*g4_7 7+0*g4_76. compute ns4_93=.9474*g4_93+.8947*g4_92+.84 21*g4_91+.7895*g4_90+.7368*g4_89+. 6842*g4_88+.6316*g4_87+ .5789*g4_86+.5263*g4_85+.4737*g4_8 4+.421*g4_83+.3684*g4_82+.3158*g4_ 81+.2632*g4_80+.2105*g4_79+ .1579*g4_78+.1053*g4_77+.0526*g4_7 6+0*g4_75. compute ns4_92=.9474*g4_92+.8947*g4_91+.84 21*g4_90+.7895*g4_89+.7368*g4_88+. 6842*g4_87+.6316*g4_86+ .5789*g4_85+.5263*g4_84+.4737*g4_8 3+.421*g4_82+.3684*g4_81+.3158*g4_ 80+.2632*g4_79+.2105*g4_78+ .1579*g4_77+.1053*g4_76+.0526*g4_7 5+0*g4_74. compute ns4_91=.9474*g4_91+.8947*g4_90+.84 21*g4_89+.7895*g4_88+.7368*g4_87+. 6842*g4_86+.6316*g4_85+ .5789*g4_84+.5263*g4_83+.4737*g4_8 2+.421*g4_81+.3684*g4_80+.3158*g4_ 79+.2632*g4_78+.2105*g4_77+ .1579*g4_76+.1053*g4_75+.0526*g4_7 4+0*g4_73. compute ns4_90=.9474*g4_90+.8947*g4_89+.84 21*g4_88+.7895*g4_87+.7368*g4_86+. 6842*g4_85+.6316*g4_84+ .5789*g4_83+.5263*g4_82+.4737*g4_8 1+.421*g4_80+.3684*g4_79+.3158*g4_ 78+.2632*g4_77+.2105*g4_76+ .1579*g4_75+.1053*g4_74+.0526*g4_7 3+0*g4_72. compute ns4_89=.9474*g4_89+.8947*g4_88+.84 21*g4_87+.7895*g4_86+.7368*g4_85+. 6842*g4_84+.6316*g4_83+ .5789*g4_82+.5263*g4_81+.4737*g4_8 0+.421*g4_79+.3684*g4_78+.3158*g4_ 77+.2632*g4_76+.2105*g4_75+ .1579*g4_74+.1053*g4_73+.0526*g4_7 2+0*g4_71. compute ns4_88=.9474*g4_88+.8947*g4_87+.84 21*g4_86+.7895*g4_85+.7368*g4_84+. 6842*g4_83+.6316*g4_82+ .5789*g4_81+.5263*g4_80+.4737*g4_7 64 9+.421*g4_78+.3684*g4_77+.3158*g4_ 76+.2632*g4_75+.2105*g4_74+ .1579*g4_73+.1053*g4_72+.0526*g4_7 1+0*g4_70. compute ns4_87=.9474*g4_87+.8947*g4_86+.84 21*g4_85+.7895*g4_84+.7368*g4_83+. 6842*g4_82+.6316*g4_81+ .5789*g4_80+.5263*g4_79+.4737*g4_7 8+.421*g4_77+.3684*g4_76+.3158*g4_ 75+.2632*g4_74+.2105*g4_73+ .1579*g4_72+.1053*g4_71+.0526*g4_7 0. compute ns4_86=.9474*g4_86+.8947*g4_85+.84 21*g4_84+.7895*g4_83+.7368*g4_82+. 6842*g4_81+.6316*g4_80+ .5789*g4_79+.5263*g4_78+.4737*g4_7 7+.421*g4_76+.3684*g4_75+.3158*g4_ 74+.2632*g4_73+.2105*g4_72+ .1579*g4_71+.1053*g4_70. compute ns4_85=.9474*g4_85+.8947*g4_84+.84 21*g4_83+.7895*g4_82+.7368*g4_81+. 6842*g4_80+.6316*g4_79+ .5789*g4_78+.5263*g4_77+.4737*g4_7 6+.421*g4_75+.3684*g4_74+.3158*g4_ 73+.2632*g4_72+.2105*g4_71+ .1579*g4_70. compute ns4_84=.9474*g4_84+.8947*g4_83+.84 21*g4_82+.7895*g4_81+.7368*g4_80+. 6842*g4_79+.6316*g4_78+ .5789*g4_77+.5263*g4_76+.4737*g4_7 5+.421*g4_74+.3684*g4_73+.3158*g4_ 72+.2632*g4_71+.2105*g4_70. compute ns4_83=.9474*g4_83+.8947*g4_82+.84 21*g4_81+.7895*g4_80+.7368*g4_79+. 6842*g4_78+.6316*g4_77+ .5789*g4_76+.5263*g4_75+.4737*g4_7 4+.421*g4_73+.3684*g4_72+.3158*g4_ 71+.2632*g4_70. compute ns4_82=.9474*g4_82+.8947*g4_81+.84 21*g4_80+.7895*g4_79+.7368*g4_78+. 6842*g4_77+.6316*g4_76+ .5789*g4_75+.5263*g4_74+.4737*g4_7 3+.421*g4_72+.3684*g4_71+.3158*g4_ 70. compute ns4_81=.9474*g4_81+.8947*g4_80+.84 21*g4_79+.7895*g4_78+.7368*g4_77+. 6842*g4_76+.6316*g4_75+ .5789*g4_74+.5263*g4_73+.4737*g4_7 2+.421*g4_71+.3684*g4_70. compute ns4_80=.9474*g4_80+.8947*g4_79+.84 21*g4_78+.7895*g4_77+.7368*g4_76+. 6842*g4_75+.6316*g4_74+ .5789*g4_73+.5263*g4_72+.4737*g4_7 1+.421*g4_70. compute ns4_79=.9474*g4_79+.8947*g4_78+.84 21*g4_77+.7895*g4_76+.7368*g4_75+. 6842*g4_74+.6316*g4_73+ .5789*g4_72+.5263*g4_71+.4737*g4_7 0. compute ns4_78=.9474*g4_78+.8947*g4_77+.84 21*g4_76+.7895*g4_75+.7368*g4_74+. 6842*g4_73+.6316*g4_72+ .5789*g4_71+.5263*g4_70. compute ns4_77=.9474*g4_77+.8947*g4_76+.84 21*g4_75+.7895*g4_74+.7368*g4_73+. 6842*g4_72+.6316*g4_71+ .5789*g4_70. compute ns4_76=.9474*g4_76+.8947*g4_75+.84 65 21*g4_74+.7895*g4_73+.7368*g4_72+. 6842*g4_71+.6316*g4_70. compute ns4_75=.9474*g4_75+.8947*g4_74+.84 21*g4_73+.7895*g4_72+.7368*g4_71+. 6842*g4_70. compute ns4_74=.9474*g4_74+.8947*g4_73+.84 21*g4_72+.7895*g4_71+.7368*g4_70. compute ns4_73=.9474*g4_73+.8947*g4_72+.84 21*g4_71+.7895*g4_70. compute ns4_72=.9474*g4_72+.8947*g4_71+.84 21*g4_70. compute ns4_71=.9474*g4_71+.8947*g4_70. compute ns4_70=.9474*g4_70. 57*g3_96+.7143*g3_95+.6429*g3_94+. 57154*g3_93+ compute ns3_01=.9286*g3_01+.8571*g3_00+.78 57*g3_99+.7143*g3_98+.6429*g3_97+. 57154*g3_96+ .5*g3_90+.4286*g3_89+.3571*g3_88+. 2857*g3_87+.2143*g3_86+.1429*g3_8 5+.0714*g3_84. compute ns3_95=.9286*g3_95+.8571*g3_94+.78 57*g3_93+.7143*g3_92+.6429*g3_91+. 57154*g3_90+ .5*g3_95+.4286*g3_94+.3571*g3_93+. 2857*g3_92+.2143*g3_91+.1429*g3_9 0+.0714*g3_89. compute ns3_00=.9286*g3_00+.8571*g3_99+.78 57*g3_98+.7143*g3_97+.6429*g3_96+. 57154*g3_95+ .5*g3_94+.4286*g3_93+.3571*g3_92+. 2857*g3_91+.2143*g3_90+.1429*g3_8 9+.0714*g3_88. compute ns3_99=.9286*g3_99+.8571*g3_98+.78 57*g3_97+.7143*g3_96+.6429*g3_95+. 57154*g3_94+ .5*g3_93+.4286*g3_92+.3571*g3_91+. 2857*g3_90+.2143*g3_89+.1429*g3_8 8+.0714*g3_87. compute ns3_98=.9286*g3_98+.8571*g3_97+.78 .5*g3_92+.4286*g3_91+.3571*g3_90+. 2857*g3_89+.2143*g3_88+.1429*g3_8 7+.0714*g3_86. compute ns3_97=.9286*g3_97+.8571*g3_96+.78 57*g3_95+.7143*g3_94+.6429*g3_93+. 57154*g3_92+ .5*g3_91+.4286*g3_90+.3571*g3_89+. 2857*g3_88+.2143*g3_87+.1429*g3_8 6+.0714*g3_85. compute ns3_96=.9286*g3_96+.8571*g3_95+.78 57*g3_94+.7143*g3_93+.6429*g3_92+. 57154*g3_91+ .5*g3_89+.4286*g3_88+.3571*g3_87+. 2857*g3_86+.2143*g3_85+.1429*g3_8 4+.0714*g3_83+0*g3_82. compute ns3_94=.9286*g3_94+.8571*g3_93+.78 57*g3_92+.7143*g3_91+.6429*g3_90+. 57154*g3_89+ .5*g3_88+.4286*g3_87+.3571*g3_86+. 2857*g3_85+.2143*g3_84+.1429*g3_8 3+.0714*g3_82+0*g3_81. compute ns3_93=.9286*g3_93+.8571*g3_92+.78 57*g3_91+.7143*g3_90+.6429*g3_89+. 57154*g3_88+ .5*g3_87+.4286*g3_86+.3571*g3_85+. 2857*g3_84+.2143*g3_83+.1429*g3_8 2+.0714*g3_81+0*g3_80. 66 compute ns3_92=.9286*g3_92+.8571*g3_91+.78 57*g3_90+.7143*g3_89+.6429*g3_88+. 57154*g3_87+ .5*g3_86+.4286*g3_85+.3571*g3_84+. 2857*g3_83+.2143*g3_82+.1429*g3_8 1+.0714*g3_80+0*g3_79. compute ns3_91=.9286*g3_91+.8571*g3_90+.78 57*g3_89+.7143*g3_88+.6429*g3_87+. 57154*g3_86+ .5*g3_85+.4286*g3_84+.3571*g3_83+. 2857*g3_82+.2143*g3_81+.1429*g3_8 0+.0714*g3_79+0*g3_78. compute ns3_90=.9286*g3_90+.8571*g3_89+.78 57*g3_88+.7143*g3_87+.6429*g3_86+. 57154*g3_85+ .5*g3_84+.4286*g3_83+.3571*g3_82+. 2857*g3_81+.2143*g3_80+.1429*g3_7 9+.0714*g3_78+0*g3_77. compute ns3_89=.9286*g3_89+.8571*g3_88+.78 57*g3_87+.7143*g3_86+.6429*g3_85+. 57154*g3_84+ .5*g3_83+.4286*g3_82+.3571*g3_81+. 2857*g3_80+.2143*g3_79+.1429*g3_7 8+.0714*g3_77+0*g3_76. compute ns3_88=.9286*g3_88+.8571*g3_87+.78 57*g3_86+.7143*g3_85+.6429*g3_84+. 57154*g3_83+ .5*g3_82+.4286*g3_81+.3571*g3_80+. 2857*g3_79+.2143*g3_78+.1429*g3_7 7+.0714*g3_76+0*g3_75. compute ns3_87=.9286*g3_87+.8571*g3_86+.78 57*g3_85+.7143*g3_84+.6429*g3_83+. 57154*g3_82+ .5*g3_81+.4286*g3_80+.3571*g3_79+. 2857*g3_78+.2143*g3_77+.1429*g3_7 6+.0714*g3_75+0*g3_74. compute ns3_86=.9286*g3_86+.8571*g3_85+.78 57*g3_84+.7143*g3_83+.6429*g3_82+. 57154*g3_81+ .5*g3_80+.4286*g3_79+.3571*g3_78+. 2857*g3_77+.2143*g3_76+.1429*g3_7 5+.0714*g3_74+0*g3_73. compute ns3_85=.9286*g3_85+.8571*g3_84+.78 57*g3_83+.7143*g3_82+.6429*g3_81+. 57154*g3_80+ .5*g3_79+.4286*g3_78+.3571*g3_77+. 2857*g3_76+.2143*g3_75+.1429*g3_7 4+.0714*g3_73+0*g3_72. compute ns3_84=.9286*g3_84+.8571*g3_83+.78 57*g3_82+.7143*g3_81+.6429*g3_80+. 57154*g3_79+ .5*g3_78+.4286*g3_77+.3571*g3_76+. 2857*g3_75+.2143*g3_74+.1429*g3_7 3+.0714*g3_72+0*g3_71. compute ns3_83=.9286*g3_83+.8571*g3_82+.78 57*g3_81+.7143*g3_80+.6429*g3_79+. 57154*g3_78+ .5*g3_77+.4286*g3_76+.3571*g3_75+. 2857*g3_74+.2143*g3_73+.1429*g3_7 2+.0714*g3_71+0*g3_70. compute ns3_82=.9286*g3_82+.8571*g3_81+.78 57*g3_80+.7143*g3_79+.6429*g3_78+. 57154*g3_77+ .5*g3_76+.4286*g3_75+.3571*g3_74+. 2857*g3_73+.2143*g3_72+.1429*g3_7 1+.0714*g3_70. compute ns3_81=.9286*g3_81+.8571*g3_80+.78 67 57*g3_79+.7143*g3_78+.6429*g3_77+. 57154*g3_76+ .5*g3_75+.4286*g3_74+.3571*g3_73+. 2857*g3_72+.2143*g3_71+.1429*g3_7 0. compute ns3_80=.9286*g3_80+.8571*g3_79+.78 57*g3_78+.7143*g3_77+.6429*g3_76+. 57154*g3_75+ .5*g3_74+.4286*g3_73+.3571*g3_72+. 2857*g3_71+.2143*g3_70. compute ns3_79=.9286*g3_79+.8571*g3_78+.78 57*g3_77+.7143*g3_76+.6429*g3_75+. 57154*g3_74+ .5*g3_73+.4286*g3_72+.3571*g3_71+. 2857*g3_70. compute ns3_78=.9286*g3_78+.8571*g3_77+.78 57*g3_76+.7143*g3_75+.6429*g3_74+. 57154*g3_73+ .5*g3_72+.4286*g3_71+.3571*g3_70. compute ns3_77=.9286*g3_77+.8571*g3_76+.78 57*g3_75+.7143*g3_74+.6429*g3_73+. 57154*g3_72+ .5*g3_71+.4286*g3_70. compute ns3_76=.9286*g3_76+.8571*g3_75+.78 57*g3_74+.7143*g3_73+.6429*g3_72+. 57154*g3_71+ .5*g3_70. compute ns3_75=.9286*g3_75+.8571*g3_74+.78 57*g3_73+.7143*g3_72+.6429*g3_71+. 57154*g3_70. compute ns3_74=.9286*g3_74+.8571*g3_73+.78 57*g3_72+.7143*g3_71+.6429*g3_70. compute ns3_73=.9286*g3_73+.8571*g3_72+.78 57*g3_71+.7143*g3_70. compute ns3_72=.9286*g3_72+.8571*g3_71+.78 57*g3_70. compute ns3_71=.9286*g3_71+.8571*g3_70. compute ns3_70=.9286*g3_70. compute ns2_01=.9167*g2_01+.833*g2_00+.75* g2_99+.6667*g2_98+.5833*g2_97+.5*g 2_96+.4167*g2_95+.3334*g2_94+ .25*g2_93+.1667*g2_92+.0833*g2_91. compute ns2_00=.9167*g2_00+.833*g2_99+.75* g2_98+.6667*g2_97+.5833*g2_96+.5*g 2_95+.4167*g2_94+.3334*g2_93+ .25*g2_92+.1667*g2_91+.0833*g2_90. compute ns2_99=.9167*g2_99+.833*g2_98+.75* g2_97+.6667*g2_96+.5833*g2_95+.5*g 2_94+.4167*g2_93+.3334*g2_92+ .25*g2_91+.1667*g2_90+.0833*g2_89. compute ns2_98=.9167*g2_98+.833*g2_97+.75* g2_96+.6667*g2_95+.5833*g2_94+.5*g 2_93+.4167*g2_92+.3334*g2_91+ .25*g2_90+.1667*g2_89+.0833*g2_88. compute ns2_97=.9167*g2_97+.833*g2_96+.75* g2_95+.6667*g2_94+.5833*g2_93+.5*g 2_92+.4167*g2_91+.3334*g2_90+ .25*g2_89+.1667*g2_88+.0833*g2_87. compute ns2_96=.9167*g2_96+.833*g2_95+.75* g2_94+.6667*g2_93+.5833*g2_92+.5*g 2_91+.4167*g2_90+.3334*g2_89+ .25*g2_88+.1667*g2_87+.0833*g2_86. compute ns2_95=.9167*g2_95+.833*g2_94+.75* 68 g2_93+.6667*g2_92+.5833*g2_91+.5*g 2_90+.4167*g2_89+.3334*g2_88+ .25*g2_87+.1667*g2_86+.0833*g2_85+ 0*g2_85. compute ns2_94=.9167*g2_94+.833*g2_93+.75* g2_92+.6667*g2_91+.5833*g2_90+.5*g 2_89+.4167*g2_88+.3334*g2_87+ .25*g2_86+.1667*g2_85+.0833*g2_84+ 0*g2_83. compute ns2_93=.9167*g2_93+.833*g2_92+.75* g2_91+.6667*g2_90+.5833*g2_89+.5*g 2_88+.4167*g2_87+.3334*g2_86+ .25*g2_85+.1667*g2_84+.0833*g2_83+ 0*g2_82. compute ns2_92=.9167*g2_92+.833*g2_91+.75* g2_90+.6667*g2_89+.5833*g2_88+.5*g 2_87+.4167*g2_86+.3334*g2_85+ .25*g2_84+.1667*g2_83+.0833*g2_82+ 0*g2_81. compute ns2_91=.9167*g2_91+.833*g2_90+.75* g2_89+.6667*g2_88+.5833*g2_87+.5*g 2_86+.4167*g2_85+.3334*g2_84+ .25*g2_83+.1667*g2_82+.0833*g2_81+ 0*g2_80. compute ns2_90=.9167*g2_90+.833*g2_89+.75* g2_88+.6667*g2_87+.5833*g2_86+.5*g 2_85+.4167*g2_84+.3334*g2_83+ .25*g2_82+.1667*g2_81+.0833*g2_80+ 0*g2_79. compute ns2_89=.9167*g2_89+.833*g2_88+.75* g2_87+.6667*g2_86+.5833*g2_85+.5*g 2_84+.4167*g2_83+.3334*g2_82+ .25*g2_81+.1667*g2_80+.0833*g2_79+ 0*g2_78. compute ns2_88=.9167*g2_88+.833*g2_87+.75* g2_86+.6667*g2_85+.5833*g2_84+.5*g 2_83+.4167*g2_82+.3334*g2_81+ .25*g2_80+.1667*g2_79+.0833*g2_78+ 0*g2_77. compute ns2_87=.9167*g2_87+.833*g2_86+.75* g2_85+.6667*g2_84+.5833*g2_83+.5*g 2_82+.4167*g2_81+.3334*g2_80+ .25*g2_79+.1667*g2_78+.0833*g2_77+ 0*g2_76. compute ns2_86=.9167*g2_86+.833*g2_85+.75* g2_84+.6667*g2_83+.5833*g2_82+.5*g 2_81+.4167*g2_80+.3334*g2_79+ .25*g2_78+.1667*g2_77+.0833*g2_76+ 0*g2_75. compute ns2_85=.9167*g2_85+.833*g2_84+.75* g2_83+.6667*g2_82+.5833*g2_81+.5*g 2_80+.4167*g2_79+.3334*g2_78+ .25*g2_77+.1667*g2_76+.0833*g2_75+ 0*g2_74. compute ns2_84=.9167*g2_84+.833*g2_83+.75* g2_82+.6667*g2_81+.5833*g2_80+.5*g 2_79+.4167*g2_78+.3334*g2_77+ .25*g2_76+.1667*g2_75+.0833*g2_74+ 0*g2_73. compute ns2_83=.9167*g2_83+.833*g2_82+.75* g2_81+.6667*g2_80+.5833*g2_79+.5*g 2_78+.4167*g2_77+.3334*g2_76+ .25*g2_75+.1667*g2_74+.0833*g2_73+ 0*g2_72. 69 compute ns2_82=.9167*g2_82+.833*g2_81+.75* g2_80+.6667*g2_79+.5833*g2_78+.5*g 2_77+.4167*g2_76+.3334*g2_75+ .25*g2_74+.1667*g2_73+.0833*g2_72+ 0*g2_71. compute ns2_81=.9167*g2_81+.833*g2_80+.75* g2_79+.6667*g2_78+.5833*g2_77+.5*g 2_76+.4167*g2_75+.3334*g2_74+ .25*g2_73+.1667*g2_72+.0833*g2_71+ 0*g2_70. compute ns2_80=.9167*g2_80+.833*g2_79+.75* g2_78+.6667*g2_77+.5833*g2_76+.5*g 2_75+.4167*g2_74+.3334*g2_73+ .25*g2_72+.1667*g2_71+.0833*g2_70. compute ns2_79=.9167*g2_79+.833*g2_78+.75* g2_77+.6667*g2_76+.5833*g2_75+.5*g 2_74+.4167*g2_73+.3334*g2_72+ .25*g2_71+.1667*g2_70. compute ns2_78=.9167*g2_78+.833*g2_77+.75* g2_76+.6667*g2_75+.5833*g2_74+.5*g 2_73+.4167*g2_72+.3334*g2_71+ .25*g2_70. compute ns2_77=.9167*g2_77+.833*g2_76+.75* g2_75+.6667*g2_74+.5833*g2_73+.5*g 2_72+.4167*g2_71+.3334*g2_70. compute ns2_76=.9167*g2_76+.833*g2_75+.75* g2_74+.6667*g2_73+.5833*g2_72+.5*g 2_71+.4167*g2_70. compute ns2_75=.9167*g2_75+.833*g2_74+.75* g2_73+.6667*g2_72+.5833*g2_71+.5*g 2_70. compute ns2_74=.9167*g2_74+.833*g2_73+.75* g2_72+.6667*g2_71+.5833*g2_70. compute ns2_73=.9167*g2_73+.833*g2_72+.75* g2_71+.6667*g2_70. compute ns2_72=.9167*g2_72+.833*g2_71+.75* g2_70. compute ns2_71=.9167*g2_71+.833*g2_70. compute ns2_70=.9167*g2_70. compute ns1_01=.8*g1_01+.6*g1_00+.4*g1_99+ .2*g1_98. compute ns1_00=.8*g1_00+.6*g1_99+.4*g1_98+ .2*g1_97. compute ns1_99=.8*g1_99+.6*g1_98+.4*g1_97+ .2*g1_96. compute ns1_98=.8*g1_98+.6*g1_97+.4*g1_96+ .2*g1_95. compute ns1_97=.8*g1_97+.6*g1_96+.4*g1_95+ .2*g1_94. compute ns1_96=.8*g1_96+.6*g1_95+.4*g1_94+ .2*g1_93. compute ns1_95=.8*g1_95+.6*g1_94+.4*g1_93+ .2*g1_92+0*g1_91. compute ns1_94=.8*g1_94+.6*g1_93+.4*g1_92+ .2*g1_91+0*g1_90. compute ns1_93=.8*g1_93+.6*g1_92+.4*g1_91+ .2*g1_90+0*g1_89. compute ns1_92=.8*g1_92+.6*g1_91+.4*g1_90+ .2*g1_89+0*g1_88. compute ns1_91=.8*g1_91+.6*g1_90+.4*g1_89+ .2*g1_88+0*g1_87. compute ns1_90=.8*g1_90+.6*g1_89+.4*g1_88+ .2*g1_87+0*g1_86. 70 compute ns1_89=.8*g1_89+.6*g1_88+.4*g1_87+ .2*g1_86+0*g1_85. compute ns1_88=.8*g1_88+.6*g1_87+.4*g1_86+ .2*g1_85+0*g1_84. compute ns1_87=.8*g1_87+.6*g1_86+.4*g1_85+ .2*g1_84+0*g1_83. compute ns1_86=.8*g1_86+.6*g1_85+.4*g1_84+ .2*g1_83+0*g1_82. compute ns1_85=.8*g1_85+.6*g1_84+.4*g1_83+ .2*g1_82+0*g1_81. compute ns1_84=.8*g1_84+.6*g1_83+.4*g1_82+ .2*g1_81+0*g1_80. compute ns1_83=.8*g1_83+.6*g1_82+.4*g1_81+ .2*g1_80+0*g1_79. compute ns1_82=.8*g1_82+.6*g1_81+.4*g1_80+ .2*g1_79+0*g1_78. compute ns1_81=.8*g1_81+.6*g1_80+.4*g1_79+ .2*g1_78+0*g1_77. compute ns1_80=.8*g1_80+.6*g1_79+.4*g1_78+ .2*g1_77+0*g1_76. compute ns1_79=.8*g1_79+.6*g1_78+.4*g1_77+ .2*g1_76+0*g1_75. compute ns1_78=.8*g1_78+.6*g1_77+.4*g1_76+ .2*g1_75+0*g1_74. compute ns1_77=.8*g1_77+.6*g1_76+.4*g1_75+ .2*g1_74+0*g1_73. compute ns1_76=.8*g1_76+.6*g1_75+.4*g1_74+ .2*g1_73+0*g1_72. compute ns1_75=.8*g1_75+.6*g1_74+.4*g1_73+ .2*g1_72+0*g1_71. compute ns1_74=.8*g1_74+.6*g1_73+.4*g1_72+ .2*g1_71+0*g1_70. compute ns1_73=.8*g1_73+.6*g1_72+.4*g1_71+ .2*g1_70. compute ns1_72=.8*g1_72+.6*g1_71+.4*g1_70. compute ns1_71=.8*g1_71+.6*g1_70. compute ns1_70=.8*g1_70. compute gs01=gs1_01+gs2_01+gs3_01+gs4_01+ gs5_01+gs6_01. compute gs00=gs1_00+gs2_00+gs3_00+gs4_00+ gs5_00+gs6_00. compute gs99=gs1_99+gs2_99+gs3_99+gs4_99+ gs5_99+gs6_99. compute gs98=gs1_98+gs2_98+gs3_98+gs4_98+ gs5_98+gs6_98. compute gs97=gs1_97+gs2_97+gs3_97+gs4_97+ gs5_97+gs6_97. compute gs96=gs1_96+gs2_96+gs3_96+gs4_96+ gs5_96+gs6_96. compute gs95=gs1_95+gs2_95+gs3_95+gs4_95+ gs5_95+gs6_95. compute gs94=gs1_94+gs2_94+gs3_94+gs4_94+ gs5_94+gs6_94. compute gs93=gs1_93+gs2_93+gs3_93+gs4_93+ gs5_93+gs6_93. compute gs92=gs1_92+gs2_92+gs3_92+gs4_92+ gs5_92+gs6_92. compute gs91=gs1_91+gs2_91+gs3_91+gs4_91+ gs5_91+gs6_91. 71 compute gs90=gs1_90+gs2_90+gs3_90+gs4_90+ gs5_90+gs6_90. compute gs89=gs1_89+gs2_89+gs3_89+gs4_89+ gs5_89+gs6_89. compute gs88=gs1_88+gs2_88+gs3_88+gs4_88+ gs5_88+gs6_88. compute gs87=gs1_87+gs2_87+gs3_87+gs4_87+ gs5_87+gs6_87. compute gs86=gs1_86+gs2_86+gs3_86+gs4_86+ gs5_86+gs6_86. compute gs85=gs1_85+gs2_85+gs3_85+gs4_85+ gs5_85+gs6_85. compute gs84=gs1_84+gs2_84+gs3_84+gs4_84+ gs5_84+gs6_84. compute gs83=gs1_83+gs2_83+gs3_83+gs4_83+ gs5_83+gs6_83. compute gs82=gs1_82+gs2_82+gs3_82+gs4_82+ gs5_82+gs6_82. compute gs81=gs1_81+gs2_81+gs3_81+gs4_81+ gs5_81+gs6_81. compute gs80=gs1_80+gs2_80+gs3_80+gs4_80+ gs5_80+gs6_80. compute gs79=gs1_79+gs2_79+gs3_79+gs4_79+ gs5_79+gs6_79. compute gs78=gs1_78+gs2_78+gs3_78+gs4_78+ gs5_78+gs6_78. compute gs77=gs1_77+gs2_77+gs3_77+gs4_77+ gs5_77+gs6_77. compute gs76=gs1_76+gs2_76+gs3_76+gs4_76+ gs5_76+gs6_76. compute gs75=gs1_75+gs2_75+gs3_75+gs4_75+ gs5_75+gs6_75. compute gs74=gs1_74+gs2_74+gs3_74+gs4_74+ gs5_74+gs6_74. compute gs73=gs1_73+gs2_73+gs3_73+gs4_73+ gs5_73+gs6_73. compute gs72=gs1_72+gs2_72+gs3_72+gs4_72+ gs5_72+gs6_72. compute gs71=gs1_71+gs2_71+gs3_71+gs4_71+ gs5_71+gs6_71. compute gs70=gs1_70+gs2_70+gs3_70+gs4_70+ gs5_70+gs6_70. compute ns01=ns1_01+ns2_01+ns3_01+ns4_01+ ns5_01+ns6_01. compute ns00=ns1_00+ns2_00+ns3_00+ns4_00+ ns5_00+ns6_00. compute ns99=ns1_99+ns2_99+ns3_99+ns4_99+ ns5_99+ns6_99. compute ns98=ns1_98+ns2_98+ns3_98+ns4_98+ ns5_98+ns6_98. compute ns97=ns1_97+ns2_97+ns3_97+ns4_97+ ns5_97+ns6_97. compute ns96=ns1_96+ns2_96+ns3_96+ns4_96+ ns5_96+ns6_96. compute ns95=ns1_95+ns2_95+ns3_95+ns4_95+ ns5_95+ns6_95. compute ns94=ns1_94+ns2_94+ns3_94+ns4_94+ ns5_94+ns6_94. compute ns93=ns1_93+ns2_93+ns3_93+ns4_93+ ns5_93+ns6_93. 72 compute ns92=ns1_92+ns2_92+ns3_92+ns4_92+ ns5_92+ns6_92. compute ns91=ns1_91+ns2_91+ns3_91+ns4_91+ ns5_91+ns6_91. compute ns90=ns1_90+ns2_90+ns3_90+ns4_90+ ns5_90+ns6_90. compute ns89=ns1_89+ns2_89+ns3_89+ns4_89+ ns5_89+ns6_89. compute ns88=ns1_88+ns2_88+ns3_88+ns4_88+ ns5_88+ns6_88. compute ns87=ns1_87+ns2_87+ns3_87+ns4_87+ ns5_87+ns6_87. compute ns86=ns1_86+ns2_86+ns3_86+ns4_86+ ns5_86+ns6_86. compute ns85=ns1_85+ns2_85+ns3_85+ns4_85+ ns5_85+ns6_85. compute ns84=ns1_84+ns2_84+ns3_84+ns4_84+ ns5_84+ns6_84. compute ns83=ns1_83+ns2_83+ns3_83+ns4_83+ ns5_83+ns6_83. compute ns82=ns1_82+ns2_82+ns3_82+ns4_82+ ns5_82+ns6_82. compute ns81=ns1_81+ns2_81+ns3_81+ns4_81+ ns5_81+ns6_81. compute ns80=ns1_80+ns2_80+ns3_80+ns4_80+ ns5_80+ns6_80. compute ns79=ns1_79+ns2_79+ns3_79+ns4_79+ ns5_79+ns6_79. compute ns78=ns1_78+ns2_78+ns3_78+ns4_78+ ns5_78+ns6_78. compute ns77=ns1_77+ns2_77+ns3_77+ns4_77+ ns5_77+ns6_77. compute ns76=ns1_76+ns2_76+ns3_76+ns4_76+ ns5_76+ns6_76. compute ns75=ns1_75+ns2_75+ns3_75+ns4_75+ ns5_75+ns6_75. compute ns74=ns1_74+ns2_74+ns3_74+ns4_74+ ns5_74+ns6_74. compute ns73=ns1_73+ns2_73+ns3_73+ns4_73+ ns5_73+ns6_73. compute ns72=ns1_72+ns2_72+ns3_72+ns4_72+ ns5_72+ns6_72. compute ns71=ns1_71+ns2_71+ns3_71+ns4_71+ ns5_71+ns6_71. compute ns70=ns1_70+ns2_70+ns3_70+ns4_70+ ns5_70+ns6_70. compute cap01=0.66*gs01+0.34*ns01. compute cap00=0.66*gs00+0.34*ns00. compute cap99=0.66*gs99+0.34*ns99. compute cap98=0.66*gs98+0.34*ns98. compute cap97=0.66*gs97+0.34*ns97. compute cap96=0.66*gs96+0.34*ns96. compute cap95=0.66*gs95+0.34*ns95. compute cap94=0.66*gs94+0.34*ns94. compute cap93=0.66*gs93+0.34*ns93. compute cap92=0.66*gs92+0.34*ns92. compute cap91=0.66*gs91+0.34*ns91. compute cap90=0.66*gs90+0.34*ns90. compute cap89=0.66*gs89+0.34*ns89. compute cap88=0.66*gs88+0.34*ns88. compute cap87=0.66*gs87+0.34*ns87. compute cap86=0.66*gs86+0.34*ns86. compute cap85=0.66*gs85+0.34*ns85. compute cap84=0.66*gs84+0.34*ns84. compute cap83=0.66*gs83+0.34*ns83. compute cap82=0.66*gs82+0.34*ns82. compute cap81=0.66*gs81+0.34*ns81. 73 compute cap80=0.66*gs80+0.34*ns80. compute cap79=0.66*gs79+0.34*ns79. compute cap78=0.66*gs78+0.34*ns78. compute cap77=0.66*gs77+0.34*ns77. compute cap76=0.66*gs76+0.34*ns76. compute cap75=0.66*gs75+0.34*ns75. compute cap74=0.66*gs74+0.34*ns74. compute cap73=0.66*gs73+0.34*ns73. compute cap72=0.66*gs72+0.34*ns72. compute cap71=0.66*gs71+0.34*ns71. compute cap70=0.66*gs70+0.34*ns70. sort cases by cso_ref. save outfile='h:\files\capital stock\flat1.sav'/keep=cso_ref gs01 to cap70. get file='h:\files\capital stock\rnet_pm_7001.sav'. *CHANGE GROUP = TO MATCH GROUP NO. AT START OF FILE. select if group eq 1. if (year eq 1993 and cso_ref gt 80000000) cso_ref=cso_ref-90000000. sort cases by cso_ref. match files file=*/table='h:\files\capital stock\flat1.sav'/by cso_ref. if (year eq 1970) K_pm=cap70. if (year eq 1971) K_pm=cap71. if (year eq 1972) K_pm=cap72. if (year eq 1973) K_pm=cap73. if (year eq 1974) K_pm=cap74. if (year eq 1975) K_pm=cap75. if (year eq 1976) K_pm=cap76. if (year eq 1977) K_pm=cap77. if (year eq 1978) K_pm=cap78. if (year eq 1979) K_pm=cap79. if (year eq 1980) K_pm=cap80. if (year eq 1981) K_pm=cap81. if (year eq 1982) K_pm=cap82. if (year eq 1983) K_pm=cap83. if (year eq 1984) K_pm=cap84. if (year eq 1985) K_pm=cap85. if (year eq 1986) K_pm=cap86. if (year eq 1987) K_pm=cap87. if (year eq 1988) K_pm=cap88. if (year eq 1989) K_pm=cap89. if (year eq 1990) K_pm=cap90. if (year eq 1991) K_pm=cap91. if (year eq 1992) K_pm=cap92. if (year eq 1993) K_pm=cap93. if (year eq 1994) K_pm=cap94. if (year eq 1995) K_pm=cap95. if (year eq 1996) K_pm=cap96. if (year eq 1997) K_pm=cap97. if (year eq 1998) K_pm=cap98. if (year eq 1999) K_pm=cap99. if (year eq 2000) K_pm=cap00. if (year eq 2001) K_pm=cap01. if (year eq 1970) nk_pm=ns70. if (year eq 1971) nk_pm=ns71. if (year eq 1972) nk_pm=ns72. if (year eq 1973) nk_pm=ns73. if (year eq 1974) nk_pm=ns74. if (year eq 1975) nk_pm=ns75. if (year eq 1976) nk_pm=ns76. if (year eq 1977) nk_pm=ns77. if (year eq 1978) nk_pm=ns78. if (year eq 1979) nk_pm=ns79. if (year eq 1980) nk_pm=ns80. if (year eq 1981) nk_pm=ns81. if (year eq 1982) nk_pm=ns82. if (year eq 1983) nk_pm=ns83. if (year eq 1984) nk_pm=ns84. if (year eq 1985) nk_pm=ns85. if (year eq 1986) nk_pm=ns86. if (year eq 1987) nk_pm=ns87. if (year eq 1988) nk_pm=ns88. if (year eq 1989) nk_pm=ns89. if (year eq 1990) nk_pm=ns90. if (year eq 1991) nk_pm=ns91. if (year eq 1992) nk_pm=ns92. if (year eq 1993) nk_pm=ns93. if (year eq 1994) nk_pm=ns94. if (year eq 1995) nk_pm=ns95. if (year eq 1996) nk_pm=ns96. if (year eq 1997) nk_pm=ns97. if (year eq 1998) nk_pm=ns98. if (year eq 1999) nk_pm=ns99. if (year eq 2000) nk_pm=ns00. if (year eq 2001) nk_pm=ns01. 74 if (year eq 1970) gk_pm=gs70. if (year eq 1971) gk_pm=gs71. if (year eq 1972) gk_pm=gs72. if (year eq 1973) gk_pm=gs73. if (year eq 1974) gk_pm=gs74. if (year eq 1975) gk_pm=gs75. if (year eq 1976) gk_pm=gs76. if (year eq 1977) gk_pm=gs77. if (year eq 1978) gk_pm=gs78. if (year eq 1979) gk_pm=gs79. if (year eq 1980) gk_pm=gs80. if (year eq 1981) gk_pm=gs81. if (year eq 1982) gk_pm=gs82. if (year eq 1983) gk_pm=gs83. if (year eq 1984) gk_pm=gs84. if (year eq 1985) gk_pm=gs85. if (year eq 1986) gk_pm=gs86. if (year eq 1987) gk_pm=gs87. if (year eq 1988) gk_pm=gs88. if (year eq 1989) gk_pm=gs89. if (year eq 1990) gk_pm=gs90. if (year eq 1991) gk_pm=gs91. if (year eq 1992) gk_pm=gs92. if (year eq 1993) gk_pm=gs93. if (year eq 1994) gk_pm=gs94. if (year eq 1995) gk_pm=gs95. if (year eq 1996) gk_pm=gs96. if (year eq 1997) gk_pm=gs97. if (year eq 1998) gk_pm=gs98. if (year eq 1999) gk_pm=gs99. if (year eq 2000) gk_pm=gs00. if (year eq 2001) gk_pm=gs01. descriptives cso_ref. means rnet_pm k_pm by year/cells=sum count. compute lcapst=lag(K_pm). if missing(lcapst) lcapst=0. compute deprec=rnet_pm-(K_pmlcapst). if (year eq 1970) deprec=.33*(rnet_pmnk_pm). * CHANGE FILE HANDLE PMCx SO x MATCHES GROUP NO. save outfile='H:\files\capital stock\pmc1.sav'/keep=cso_ref year rnet_pm K_pm nk_pm gk_pm deprec. get file='H:\files\capital stock\pmc1.sav'. 75 capital stock final merge.sps set mxmemory=2097151. * MERGE SUB-GROUP FILES. get file='h:\files\capital stock\pmc1.sav'. add files file=* /file='h:\files\capital stock\pmc2.sav' /file='h:\files\capital stock\pmc3.sav' /file='h:\files\capital stock\pmc4.sav' /file='h:\files\capital stock\pmc5.sav' /file='h:\files\capital stock\pmc6.sav' /file='h:\files\capital stock\pmc7.sav' /file='h:\files\capital stock\pmc8.sav' /file='h:\files\capital stock\pmc9.sav' /file='h:\files\capital stock\pmc10.sav' /file='h:\files\capital stock\pmc11.sav' /file='h:\files\capital stock\pmc12.sav' /file='h:\files\capital stock\pmc13.sav' /file='h:\files\capital stock\pmc14.sav' /file='h:\files\capital stock\pmc15.sav' /file='h:\files\capital stock\pmc16.sav' /file='h:\files\capital stock\pmc17.sav' /file='h:\files\capital stock\pmc18.sav' /file='h:\files\capital stock\pmc19.sav'/file='h:\files\capital stock\pmc20.sav'/file='h:\files\capital stock\pmc21.sav'/file='h:\files\capital stock\pmc22.sav'. aggregate outfile=* /break=cso_ref year /rnet_pm=max(rnet_pm) / K_pm=max(K_pm) / nk_pm=max(nk_pm) / gk_pm =max(gk_pm) /deprec=max(deprec). save outfile='h:\files\capital stock\net plant capital stock.sav'. if (missing(nk_69)) nk_69=0. if (missing(gk_69)) gk_69=0. if (missing(dep_69)) dep_69=0. compute k_tot=(k_pm/1000000)+k_69. compute nk_tot=nk_69+(nk_pm/1000000). compute gk_tot=gk_69+(gk_pm/1000000). compute dep_tot=dep_69+(deprec/1000000). means k_tot k_pm k_69 by year/cells=sum. aggregate outfile=* /break=cso_ref year /rnet_pm=max(rnet_pm) /k_tot=max(k_tot) /nk_tot=max(nk_tot) /gk_tot=max(gk_tot) /dep_tot=max(dep_tot). variable labels rnet_pm'real investment in plant £1980 prices' k_tot'p&m capital stock £m 1980 prices' nk_tot'net p&m capital stock £m 1980 prices' gk_tot'gross p&m capital stock £m 1980 prices' dep_tot'depreciation £m 1980 prices'. save outfile='h:\files\capital stock\total capital stocks 1970_01.sav'/keep=cso_ref year rnet_pm k_tot nk_tot gk_tot dep_tot. means k_tot rnet_pm by year/cells=sum count. title 'capital stock final merge'. get file='h:\files\capital stock\net plant capital stock.sav'. *sort cases by cso_ref year. * ADD BACK PRE-1970 BENCHMARK INFORMATION. match files file=*/table='h:\files\capital stock\benchmark.sav'/by cso_ref year. if (missing(k_69)) k_69=0. 76
© Copyright 2025 Paperzz