OECD Organisation for Economic Co-operation and Development Organisation de Coopération et de Développement Economiques OCDE STATISTICS DIRECTORATE National Accounts Division SECOND MEETING OF THE CANBERRA GROUP ON CAPITAL STOCK STATISTICS Château de la Muette, Paris 29 September - 1 October 1998 Beginning at 10 a.m. on the first day Agenda item : 6a Document number : 14 Title : The Role of Financial Capital in Production Author(s) : Steven KEUNING - Central Bureau of Statistics, Netherlands Statistics Netherlands National Accounts Department P.O. Box 4000, 2270 JM Voorburg The Netherlands THE ROLE OF FINANCIAL CAPITAL IN PRODUCTION Steven J. Keuning *) *) Head, National Accounts Department The views expressed in this paper are those of the author and do not necessarily reflect the views of Statistics Netherlands June 1998 Abstract It is increasingly acknowledged that the financial structure of a firm is an important determinant of its production costs. This paper argues that the use of a firm’s liabilities should be seen as a separate input in the production process. At the same time, the input of non-financial assets is limited to the value that is used up during the reference period. The paper elaborates on these ideas and their operationalisation in empirical work. It is concluded that the approach set out in this paper establishes a much closer relationship of general economic accounting and analysis to business economics. Table of Contents 1. Introduction................................................................................................................... 2 2. What is the capital input in production?.......................................................................... 2 3. A comparison of new and conventional estimates of multi-factor productivity change.... 9 4. Conclusions................................................................................................................. 11 List of references............................................................................................................. 12 1 1. Introduction Both conventional and modern theories of economic growth view that the contribution of financial capital to production should be analysed separately from the contribution of the ‘real’ production factors labour, physical capital, etc.1 In other words, it is assumed that the way in which a firm is financed has nothing to do with its production ‘technology’. It is increasingly acknowledged, though, that in practice the financial structure of the firm is an important determinant of its economic activity (e.g. Gertler, 1988). In any case, transactions money is needed for working capital requirements and many producers are faced with constraints on borrowing, for instance in the absence of sufficient collateral. Moreover, the cost of borrowing differs substantially among firms, countries and periods. Simultaneously, on the one hand many firms do not have access to equity capital, while on the other hand bank loans are highly non-marketable, and cannot be seen as perfect substitutes for equity. Finally, various studies point to the imperfect substitutability of financial claims to productive assets in different countries (e.g. Bovenberg and Goulder, 1991). By now, the influence of imperfect capital markets on investment decisions has been substantiated empirically in various studies (e.g. van Ees et al., 1996). However, if that is the case, the use of financial capital should also be seen as an input in the production function. This is elaborated in the next section of this paper, according to the ideas set out earlier by the author (Keuning, 1995, 1996: Section II.3.3). The third section then summarises the results from a recent study on the estimation of multi-factor productivity change (Keuning and Reininga, 1997), using both the conventional approach and the approach advocated in this paper. It appears that the incorporation of financial inputs in productivity calculations indeed throws a new light on inter-industry variations in productivity growth. This paper then winds up with some conclusions. 2. What is the capital input in production? In productivity change calculations, capital input is often equated with the use of tangible assets, such as land, machinery and buildings. Concomitantly, the physical contribution of capital to output is emphasised. This study takes the economic cost of capital inputs as a point of departure. Cost is what matters in the real world, both to the users of capital services and to the suppliers, as it represents their remuneration. Only when it is known what kind of payments or reservations have actually been made in connection with the use of capital, the underlying volume and price changes of these transactions or reservations can be disentangled. Besides, such a procedure ensures consistency with the exchange value approach that is followed in every economy-wide analysis. Ex post, the cost of capital inputs at the industry level is embodied in the gross operating surplus/mixed income generated by the industries concerned. In fact, if the (imputed) cost of selfemployed labour input is isolated from this balancing item, an estimate for the ‘pure’ capital input cost by industry remains. The next question is: what kind of capital inputs have been remunerated from this ‘residual’? Three categories of capital can be distinguished (United Nations et al., 1993: Annex V.D): 1. produced assets, consisting of fixed assets (e.g. buildings, machinery and software), inventories, and such; 1 The definition of the capital input in production has been discussed by many eminent economists in the past. Just like Triplett (1996) and Hulten (1996), this paper attempts to reconcile the capital concept that can be used for production analysis and productivity measurement with the capital concept reflected in the national (income and wealth) accounts. It is this paper’s intention, though, to add a new perspective to this long-standing debate. 2 2. non-produced, non-financial assets, such as land, subsoil assets, patented entities and purchased goodwill; and 3. financial assets/liabilities, such as currency, deposits, securities and loans. 2 For the production activities of enterprises, the following balance sheet can be drawn up: Table 1: Balance sheet for the production activities of enterprises Assets Produced assets fixed assets (e.g. buildings, machinery & equipment, software) inventories (e.g. supplies, finished goods, goods for resale) Non-produced, non-financial assets (e.g. land, patents, goodwill) Financial assets (currency, non-interest bearing deposits) Liabilities and net worth Bonds, loans, deposits, trade credits, etc. short-term long-term Shares, other equity and net worth This balance sheet is connected to the following statement of current income and outlays from production:3 Table 2: Current income and outlays of the production activities of enterprises Income Outlays Output Intermediate consumption Consumption of (owned) fixed assets (Unpaid) Compensation of labour input Rent (hire of non-produced, non-financial assets) Consumption of (owned) non-produced, non-financial assets Interest (on bonds, loans, deposits, trade credits, etc.) Dividends and net saving This income and outlays statement can be formalised in the following equation: ~ (1) j j+ pq = px pkδkKk + wll l + hmbm + πmδmΚm + rnCn + iV ∑ j ∑ k ∑ l ∑ m ∑ ∑ m n 4 with: ~ p = (producers’) price of the output; q = volume of the output; pj = purchasers’ price of intermediate input j; xj = volume of intermediate input j;5 2 All assets and liabilities are valued at current market values. Interest bearing financial assets and equity (in other enterprises) owned by the enterprises are excluded from this balance sheet, because these assets are not directly used in the production of goods and services. The same applies to interest bearing liabilities which are used to finance such assets that are not used in production. The net worth used in production is then equal to the total market value of the assets used in production minus the total market value of the liabilities used in production. Net worth can be negative, e.g. if stock markets anticipate a future rise in corporate profits or in the value of non-produced nonfinancial assets (e.g. patents), so that present share prices are already higher than is warranted by the balance sheet statement. Equity and net worth have been combined in a single category, also because the remuneration (i.e. the production costs) for these categories cannot usually be separated (cf. section 3 below). 3 This statement combines the production account and the income distribution and use accounts of the national accounts. All outlays are valued at ‘purchasers’ prices’, that is, including taxes and such. For instance, taxes on corporate profits are included in the item ‘dividends and net saving’. The design of the tax system is obviously an important determinant of the actual composition of the balance sheet and the concomitant outlays, but an analysis of those causalities goes beyond the subject of this paper. 4 In empirical applications, ‘volume’ means constant price value. 3 pk = purchasers’ price of fixed asset k; δk = depreciation rate of fixed asset k; Kk = volume of the stock (owned) of fixed asset k; wl = (imputed) wage rate of labour type l (incl. self-employed and unpaid family workers); 6 ll = input volume of labour type l (incl. self-employed and unpaid family workers); hm = price of hiring non-produced non-financial asset type m; bm = volume of hiring non-produced non-financial asset type m; πm = purchasers’ price of non-produced non-financial asset m; δm = depreciation rate of non-produced non-financial asset m; Κm = volume of the stock (owned) of non-produced non-financial asset m; rn = price of using category n of interest bearing liabilities (bonds, loans, etc.);7 Cn = (constant price) stock value of category n of bonds, loans, etc. on the balance sheet; i = (residual) rate of return on the use of shares, other equity and net worth; and V = (constant price) stock value of the shares, other equity and net worth on the balance sheet. This equation can be viewed as the outcome of the production function of the enterprises. In other words, the enterprises have combined intermediate inputs, fixed assets, labour, non-produced nonfinancial assets and all kinds of funds (deposits, loans, equity) to generate output. Obviously, the economic production function of enterprises is more comprehensive than ‘technology’ in an engineering sense. Without access to finance no production whatsoever can take place. The role of finance as an input in production can also be illustrated by comparing two enterprises with the same output and the same production technology in an engineering sense. However, enterprise A owns its office whereas enterprise B rents it. In these circumstances, enterprise A uses both the office itself and the funds tied up in its stock value. Note that the selling price of the office determines the price of using up the office (the second term on the right-hand side of equation (1) above), but that in addition this selling price, together with the prices and volumes of all other assets of A, determines the total stock value of A’s assets, which in turn determines the volume of the funds used (i.e. tied up) in A’s production process. The average price for the use of these funds depends on the way the enterprise is financed, as is reflected on the liabilities side of A’s balance sheet, and on the prices (various interest rates, rate of return on the use of equity) that must be paid for the use of all these types of liabilities. In comparison with enterprise A, less funds are tied up in the production process of enterprise B. Enterprise B namely just uses the amount of funds needed to pay the office rent to the leasing company. On the other hand, the office rent that B pays must be higher than the office depreciation cost (consumption of fixed capital), which equals A’s cost of using up the office. If finance didn’t matter in the production function, B’s total production costs would always be higher than those of A. However, in reality this may not be the case, as B’s financing costs are lower. The exact size of B’s financing costs depends on the composition of the liabilities’ side of its balance sheet. 5 Intermediate input costs include the rent of produced assets (buildings, machinery and equipment, software, etc.). 6 If labour input is subdivided by educational level, human capital input is implicitly taken into account in the production function. 7 This price equals the nominal interest rate plus the inflation compensation on the principal. This is elaborated below. The interest rate on bank deposits equals the full rate and thus not just some sort of ‘reference’ rate (that is, the interest rate minus the so-called service charge of the bank). It follows that the bank service charge should not be treated as an intermediate input of the enterprises. 4 This example also demonstrates that non-financial assets owned by the enterprise and used for production are in fact used in two respects: first, these assets are (gradually) used up and secondly, their stock value is tied up, and thus used, in that process.8 Both these aspects should be reflected in the production function. At present, mainstream theory and empirical research in general economics do not distinguish financial capital as a separate factor of production. Harper (1997) concludes “By excluding financial assets, productivity economists have effectively classified portfolio management decisions as “investor” issues. These issues presumably are not of concern to our stylized “production manager”. ... Miller and Modigliani (1966) discussed the conditions under which production decisions could be presumed separable from decisions affecting the firms financial portfolio. It is clear that most of the productivity literature has assumed these decisions are separable.” However, these conditions, including the absence of distortions by taxation, low bankruptcy costs, perfect(ly symmetric) information, fully competitive markets, are often not fulfilled in practice. Moreover, even if financing and ‘production’ decisions are taken separately, still the economic performance of firms, industries and complete economies is highly dependent upon their financing structure. The recent crisis in Asia already testifies this. Yet, productivity analyses and production functions typically define the capital input volume as the constant price value of the stock of, or services from, non-financial (fixed) assets.9 This amounts to the following equation: ~ j j+ (2) pq = px pk (δk + ρ)Kk + wll l ∑ j ∑ k ∑ l with: ρ = rate of return on the stock of fixed assets. In equation (2) a remuneration for the use of (part of) the assets side of the balance sheet is included ( ∑ pkρKk ), whereas equation (1) incorporates a remuneration for the use of the liabilities k side of the balance sheet; cf. the last two terms of equation (1) and Table 1 above. Equations (1) and (2) are more similar than they might seem at first sight, because both sides of the balance sheet obviously add up to the same total. Besides, the fourth and the fifth term on the right-hand side of equation (1) could just as well be added to the right-hand side of equation (2). Even in that case, though, the conventional equation, (2), suffers from two important shortcomings: 1) It only includes a remuneration for the use of fixed assets; in other words, it is assumed that the funds used in production are fully embodied in fixed assets, and 2) It assumes that the price change for the use of these funds depends on the price change of the non-financial (fixed) assets utilised in production, and not on the price change for the use of all kinds of liabilities (and net worth) of which these funds consist. Of course, in many applications the former deficiency is mitigated by including a remuneration for the use of land and several other non-produced non-financial assets such as patents. In terms of formula (2) above, this means that ρ covers the rate of return on a) the stock of fixed assets, plus b) the stock of non-produced non-financial assets. However, even in that case, the use of some assets is overlooked. The balance sheet of Table 1 shows that every production process also uses inventories 8 Using up an asset reflects both the decline in productive efficiency of the fixed asset because of ageing, if applicable (that is, its ‘physical’ deterioration including normal damage), ánd the (certain) decline in value because of the reduction in the remaining number of years that it can be used (cf. Triplett, 1996; section II and A.5). For any category of assets, their depreciation includes the value of the assets that are discarded under normal circumstances, including losses as a consequence of accidental damage (which is usually equal to the proportion of damage in the economy as a whole). 9 Refer to e.g. Baumol et al. (1989), Englander and Mittelstädt (1988), Jorgenson (1990), Maddison (1987), Rymes (1983) and Scott (1993). Examples of attempts to distinguish financial capital as a separate factor of production can be found in Hasan and Mahmud (1993), Stiglitz (1992) and Yeager (1979). 5 and non-interest bearing financial assets (‘cash’), in other words: working capital. Especially in activities like trade, working capital is an important factor of production. The use of working capital also reflects the fact that before someone can start producing, a fund must be available to pay for the wages, the intermediate inputs, the cost-of-living of the producer, etc. The use of this fund, which amounts to abstaining from consumption, obviously fetches a price.10 Note that the costs for using this fund are in addition to the costs of using up the intermediate inputs, fixed assets, labour services, etc. So, including these fund use costs does not involve double-counting. Even if the use of all productive assets (including working capital) was covered in the production function (2), its decomposition of changes in the value of this usage into price and volume changes is not in conformity with reality. At the end of the day, the owners of the liabilities (and net worth) are paid, not the assets. These payments reflect a) a compensation for the loss in the purchasing power of the underlying value as a consequence of inflation, b) a remuneration for the risk of bankruptcy or any other loss in the underlying value, and c) a remuneration for the owners’ willingness to abstain from consumption during the reference period. Liabilities differ particularly in the role that is played by determinants a) and b) above. Changes in the price of using the fund are determined by shifts in these three determinants, and not by a price change of the fixed assets. This is easily demonstrated for the case of a loan. If a large proportion of the liabilities-side of a firm’s balance sheet consists of short-term loans, an increase in the short-term interest rate implies a significant rise in production costs, regardless of the types of non-financial (fixed) assets used in the production process. This rise in costs is not faced by a firm with the same kind of non-financial assets in the same industry which is financed by e.g. long-term loans that do not expire in the immediate future. In addition, even though the price for the use of equity is not fixed ex ante, these funds are continuously seeking for a use with the highest expected remuneration. Take a situation where profits are expected to rise while the interest rate does not change. If under these circumstances two firms are equal except for the composition of their balance sheet, the firm that is financed to a larger extent by equity capital must realise a higher rise in operating surplus in order to offer the same rate of return growth to the investor. Again, it is the difference in the composition of the balance sheet that determines the relative rise in production costs of either firm. On the other hand price changes of fixed assets are not relevant to the whole asset stock value as is assumed in equation (2), but only to the difference between the market value of that asset at the beginning and at the end of the year, after a correction for the asset’s revaluation (see below). For, at any time, most fixed assets can be sold, and leased back if so required.11 In other words, the capital fund is not sunk in the fixed assets, but in the underlying liabilities and net worth. The first conclusion is that all production processes use a fund (equal to the total value of either side of the balance sheet) in addition to the input of labour, intermediate inputs, fixed assets, etc. The second conclusion is that the creditors supply this fund and must be paid afterwards. Thus, in order to determine the price (change) for the use of this fund, it is the composition of the liabilities side of the balance sheet that matters and not the composition of the assets side. The third, related conclusion is that not the stock value of (non-financial) assets, but only the (gradual) consumption of this value (that is, the reduction in this stock value), is an input in the production process. For a further operationalisation of the non-financial capital input in production, a distinction should be made between fixed assets, inventories, and non-produced, non-financial assets. When 10 In addition, in circumstances of substantial inflation it is advisable to include ‘depreciation of non-interest bearing financial assets’ as a cost item on the right-hand side of Table 2 and equation (1). This depreciation, that is, using up the financial assets concerned, is then computed as the loss in purchasing power of the stock value of non-interest bearing financial assets as a consequence of inflation. 11 In particular, this applies to economies with a well-developed lease-industry. For that reason, this statement may have been less valid several decades ago. 6 computing the input cost of using up fixed assets, it should be realised that a change in the market price of any fixed asset over time reflects a combination of two (usually contrary) price movements: a) depreciation, that is the price decrease because of ageing, and b) revaluation, that is the price change of asset types of a certain fixed age. The revaluation is not related to the production process; that is, it is excluded from the items in Table 2 and equation (1), and booked on the ‘other changes in assets’ account instead. The depreciation is thus roughly equal to the percentage difference, in the reference year, between the price of the assets concerned and the price of the same assets which are one year older. Revaluation is then the price change of an asset type of a certain fixed age (cf. Hulten, 1996: 155). However, if the price change during the reference year of a certain asset type is lower as the vintage is older, this means that economic obsolescence occurs; in that case, the price change of the new asset (corrected for quality change!) equals the revaluation for all vintages and the depreciation value can be computed as a residual. A decomposition of the change in the depreciation value between two consecutive years into a volume and price change is straight-forward, cf. formulas (1) and (2); the price change equals the price change of the new assets of the type concerned (again, corrected for quality change), and the volume change can be computed residually. The input cost of using up inventories equals the reduction in the market value of the stock during the reference year, after a correction for revaluation. For materials and supplies the gross reduction in inventories is incorporated in intermediate input costs in the national accounts. The change in the stock value of work in progress, of finished goods, or of goods for resale is already accounted for in the output value of the product group concerned; that is why the output and not the sales value is taken as the production value. Summarising, the input costs of a change in all kinds of inventories has already been included elsewhere in the system, and are thus not reflected separately on the right-hand side of Table 2. This also means that, when it comes to a decomposition of the value change into a price and volume change, the intermediate input or output price change of the product group concerned also applies to the change in inventories. Concerning the input cost of hiring and using up non-financial, non-produced assets, a distinction must first be made between the use of hired assets and the use of own assets; cf. Table 2 and equation (1). The cost of rental services associated with hiring non-financial assets (land, subsoil assets, etc.) equals the actual rents paid. The decomposition of input cost changes into price and volume changes should be done according to the same method as is applied for the (intermediate) input cost of renting produced assets. Concerning the input cost of using up owned non-financial assets the same rule applies as for fixed assets: the input cost equals the reduction in their stock value, after a correction for the revaluation of the assets concerned. The input volume equals the reduction in their stock volume or the reduction in their constant price stock value, so that the input price can then be derived. This applies to land, subsoil assets and other non-produced, non-financial assets such as patented entities and purchased goodwill. Note that the input cost (‘depreciation’) of (own) land that is not overexploited is normally equal to zero. Of course, the value of the land owned does influence the input costs indirectly, since it affects the volume of funds tied up in the production process (the financial capital input). The input costs of e.g. patents are normally positive, because their value continuously declines (apart from their revaluation) with the nearing of their expiration date. Although estimating the input volume of subsoil assets and other natural resources used up may not be too difficult in most cases (e.g. the number of barrels of oil extracted), establishing their price is more cumbersome. Of course, the output price for crude oil and such cannot be used, as it also covers all other costs (including financing costs!) made by the extractors. Even taking the quotient of the extraction activity’s net operating surplus and the output volume as the ‘pure’ resource price is clearly incorrect, because the financing costs are then overlooked. In addition, subsoil assets and other natural resources are often not explicitly recorded on the owners’ balance sheets, let alone that these statements allow for a reliable computation of the value reduction as a consequence of the 7 stock depletion. The input cost for using up these assets must therefore be estimated indirectly. For instance, in various countries the extractors’ profits are liable to a special tax or, if the extractor is state-owned, to an extraordinary dividend payment. If the rate of this special tax or dividend payment is the result of prolonged negotiations between the government and the extractor, it can be argued that the eventual rate is such that the extractor is precisely left with a ‘normal’ rate of return on his investment. The implicit costs of using up the subsoil assets are then equal to the special tax 12 or dividend receipts; as the volume used up is known, the price can thus be computed residually. Finally, the identification of the input cost of using liabilities and net worth proceeds in stages. As usual, the first step is a breakdown into categories. For instance, interest payments are costs for the use of all kinds of loans, securities other than shares, and other credits. These categories of liabilities should be subdivided, by term-structure, by type of conditionality, etc. Changes in these payments depend on changes in the principal and on changes in the interest rate (e.g. when a loan is renewed). The net operating surplus that remains after subtraction of the input cost of self-employed labour, non-produced non-financial assets and interest bearing liabilities reflects the remuneration for the use of the firm’s equity and net worth in production. The next step is a decomposition of the input value change into a price and volume change. The general rule applies that the volume change of the use of a liability equals the volume change of the principal of that liability.13 This implies that the price index for using the liabilities equals the price index of the principal times the remuneration rate index for the liability concerned. For example, in the case of a loan the remuneration rate equals the nominal interest rate, so that the remuneration rate index equals the (percentage) change in the nominal interest rate. Note that the price index of the principal usually reflects the relevant inflation rate while the remuneration rate index normally reflects the change in that inflation rate. This decomposition of changes in the value of liability use into price and volume changes is also similar to the case of a car rental. If car prices rise, so do the car rental prices, irrespective of the price development of the hiring service itself. By way of explanation, consider also the case of a firm where the price of all (non-financial) inputs and output(s) rises with the overall inflation rate, while the nominal interest rate does not change. Besides, all (non-financial) input and output volumes remain the same. In that case, the value of the financial inputs into the production process must also rise with the inflation rate. Obviously, the productivity of this firm does not change, so that the financial input volume should also remain the same. The volume change of the ‘dividends and net saving’ also equals the volume change of the underlying principal, that is the real change in the stock of ‘shares, other equity and net worth’. This real change is computed by deflating the value change on the balance sheet by a relevant inflation rate. The price change is residual by definition. This procedure implies that a real holding gain on an asset used in production leads to a higher real net worth of the enterprise and thus to a volume increase of the use of net worth in production. This is a correct interpretation because in that case relatively more funds are tied up in this production process and this greater use of inputs implies a productivity loss, ceteris paribus. When applying the above line of reasoning to an empirical analysis, operating surplus/mixed income by industry must be decomposed into the remunerations for the different types of capital inputs. Concerning actual payments for capital inputs (interest, dividends, land rents, subsoil asset rents), the main difficulty is the re-allocation of such payments by institutional (sub)sector to the industries concerned. Concerning the imputed payments for using own-account capital inputs, the construction of balance sheets by industry is indispensable. This brings us to the following observation. 12 If such a special tax or dividend payment does not exist and if the investors and the industry concerned do not worry about the (complete) depletion of the resource, the actual costs of using up the resource may have been close to nil. In that case, there must have been a downward pressure on the output price of the extracted resource until the industry profits just reflect the required rate of return on investment. 13 I would like to thank André Vanoli for pointing this out to me. 8 In the national accounts for production, the institutional units (enterprises) should be classified into more homogeneous categories than the subsectors distinguished in the present System of National Accounts (United Nations et al., 1993). For instance, non-financial corporations should be cross-classified by ownership (national private, public or foreign) ánd by principal production activity. For those categories it should then be possible to decompose changes in all input costs into price and volume changes. In fact, in modern economies the production function may be more homogeneous among firms with a similar institutional structure (e.g. multinationals versus the selfemployed) and a roughly equal type of market (e.g. fast moving consumer goods like food, detergents and cosmetics) than among all establishments in a 2- or 3-digit ISIC-category. This notion, however, leads to a radically different way of classifying production processes in the national accounts and in productivity analyses and such. It should not come as a surprise that presently available data by industry in the national accounts do not yet allow a rigorous empirical analysis of the above ideas. Yet, a first attempt to incorporate this new view on capital inputs into productivity analysis is reported next. 3. 14 A comparison of new and conventional estimates of multi-factor productivity change The new method of estimating multi-factor productivity change differs from the traditional method by the recognition of financial capital use as a separate input. This requires the compilation of balance sheets by industry. For the time being, the necessary data could only be compiled for four broad industry clusters in the Netherlands: 1. fixed capital intensive manufacturing (petroleum, other chemicals and transport equipment); 2. less fixed capital intensive manufacturing (all other manufacturing); 3. trade, hotels, restaurants and consumer goods repair services; and 4. transport, storage and communication services. The estimation of capital input price and volume changes in these industries has proceeded in stages: 15 1. Annual gross operating surplus/mixed income by industry has been split into: a. consumption of fixed capital,16 b. short-term interest payments, c. long-term interest payments, and d. dividends and retained earnings.17 2. The volume growth of the fixed assets input equals the growth rate of the consumption of fixed capital at constant prices. The fixed assets capital input weight agrees with the average share of fixed capital consumption in total output. 3. Annual opening and closing balance sheets by industry have been compiled, subdividing the liabilities/net worth side into: a. short-term deposits, securities other than shares, loans and other accounts payable; b. long-term deposits, securities other than shares, loans and other accounts payable; c. shares, other equity and net worth. The balance sheet for each year has been estimated as the average of the opening and closing balance sheet. The remuneration for using each of these categories of liabilities/net worth equals the categories b., c. and d. that were distinguished in step 1. 4. The liabilities and net worth of year t+1 at prices of year t have been estimated by deflating with the industry output price index, as this may be an ‘inflation rate’ that is relevant to the industry 14 Refer to Keuning and Reininga (1997) for a more extensive review of the compilation method and of the results. 15 An imputed compensation for self-employed labour has been subtracted first. Rents and the consumption of owned non-produced, non-financial assets has been considered negligible in the industries concerned. 16 In a more detailed approach, a subdivision by type of fixed asset would be advisable. 17 Distinguishing between dividends and retained earnings has been abandoned, because, for fiscal reasons, a large but fluctuating part of the shareholders’ remuneration consists of an increase in the market value of their shares induced, among other things, by high retained earnings. 9 concerned.18 This yields the annual volume growth of the principal of each category of liabilities/net worth. 5. The volume growth of the input of the three categories of liabilities/net worth equals the volume growth of the principal. The price changes follow from these volume changes and the value changes estimated in step 1. The input weight of the three categories of liabilities/net worth equals the average share of the remuneration categories b., c. and d. in step 1 in total output. The main difference between both methods is thus that in the new method the capital input weight is split into four categories of capital inputs, among which three categories of financial capital inputs and that the decomposition of the value change into price and volume changes diverges for each of these categories. Table 3 summarises multi-factor productivity growth rates for the four above-mentioned industry clusters: Table 3: Multi-factor productivity change (1988-1992) in the Netherlands logarithmic growth rates (%) 1991 1992 ‘88-‘92a) Chemical Industry, Petroleum Industry, and Transport Equipment Industry traditional method 0.88 0.32 -0.58 new method 1.36 0.55 -0.74 -0.09 -0.42 0.53 0.75 Other Manufacturing Industry traditional method new method 0.13 0.23 -0.45 -0.60 0.94 1.29 Trade, Hotels, Café’s, Restaurants, and Repair of Consumer Goods traditional method 0.48 0.42 new method 1.84 1.76 0.02 1.57 -1.27 -2.47 -0.35 2.75 Transport, Storage and Communication traditional method new method 2.09 2.79 0.92 1.41 7.11 5.25 years 1989 0.80 1.22 2.04 -1.26 1990 0.45 0.43 1.83 2.17 a) Cumulative growth rate for the period 1988-1992 as a whole. For both clusters of manufacturing industries, the results of the new approach resemble those of the traditional method. The productivity growth pattern is the same, by and large: a worsening of productivity change over the period 1988-1992. The productivity change estimates for the period as a whole do not differ very much either. However, the traditional method somewhat underestimates the actual productivity changes, both the positive and the negative ones. In the trade and related services industry, the new method yields quite different insights. In fact, during the period concerned a substantial productivity growth occurred, while the conventional 18 An alternative possibility would have been the average price rise of the asset-side of the balance sheet. Besides, it had been assumed that financial capital is quite mobile across industries, a general inflation rate would have been selected instead. In that case, though, it would still have been assumed that financial capital is not internationally mobile. 10 computation method yielded a small decline. Notably, a relatively large volume decrease of the input of both long-term loans, etc. and shares plus net worth contributed to the relatively high productivity growth in 1989-1991. The reverse holds for 1992. Both methods yield a declining pattern of multifactor productivity change over the period 1989-1992. Finally, in the transport, storage and communication industry the new approach results in a lower productivity growth estimate for the period as a whole. For each of the years 1990, 1991 and 1992, the outcomes of both methods are similar, albeit that the new method yields slightly higher growth rates. In 1989, however, a very high volume growth of the input of both long-term loans, etc. and shares plus net worth caused a productivity decline in this industry, which was not picked up by the traditional method. Overall, the spread of productivity change estimates among industries was, for the period as a whole, smaller according to the new method (between 0.75% and 5.25%) than according to the traditional method (between-0.35% and 7.11%). 4. Conclusions The growing awareness that the financial structure of a firm affects its activity should also be reflected in production functions, productivity calculations and so on. This paper has attempted to design and operationalise a macro-economic measurement of financial capital inputs in production and to show the consequences for the estimation of multi-factor productivity change by industry. In essence, our findings are the following. First, the input of non-financial assets in production does not differ fundamentally from intermediate inputs, albeit that the services from these assets are spread out over more than one year.19 Secondly, in addition to the costs due to the gradual consumption of non-financial assets, there are costs connected with the use of the funds tied up in these assets. These funds are used for production and cannot simultaneously be used for other purposes, such as the immediate satisfaction of wants. That abstinence must be remunerated. The essence of the argument developed here is that this remuneration should not be assigned to the kinds of assets and working capital financed with these funds, but to the categories of liabilities and net worth that acquire this income. In comparison with present macro-economic theory and practice, this implies a shift in emphasis from the assets-side of the balance sheet to the liabilities-side. The total value of both sides of the balance sheet is of course the same. What differs is the classification of items and, particularly, the decomposition of value changes into volume changes and price changes when it comes to productivity analyses, production functions, etc. A rigorous empirical application of these ideas requires a different meso- and macro-economic data base than is presently available. Following Keuning and Reininga (1997), section 3 of this paper contains an application of this new concept of capital input in production to the estimation of productivity change for four industry clusters in the Netherlands. The results have been compared with the outcomes according to the traditional estimation method. The new method has resulted in a substantially smaller range of productivity change estimates by industry. Particularly in the trade and related services industry, the new method yielded quite different results. Whereas originally this seemed the only industry with a productivity loss over the whole period concerned, the new method yielded a productivity gain, in between the gain of the manufacturing and the transport industries. This modest experiment cannot yet substantiate the accuracy or the relevance of this new approach to measuring capital input in production. Besides, more research is needed into the most accurate deflator for various categories of liabilities under various circumstances; cf. foot note 18. However, it may be worth-while to repeat this exercise for other countries and other periods. In addition, it may be of interest to compare the productivity performance of countries, or individual firms, with this new method. 19 In so far as well-developed markets for second-hand capital goods do not exist, these commodities are less fungible than intermediate inputs. However, the delivery of intermediate inputs may also be fixed in long-term contracts. 11 By recasting the model of economic production in the way described here, a new light is thrown on differences in productivity growth among firms, industries or countries. In turn, that may yield a new perspective on the determinants of economic growth. Finally, the approach in this study also establishes a much closer link of macro-economic accounting and analysis to business economics. In fact, the relative neglect of financial inputs in present mainstream macro-economic theory of production and growth is all the more surprising, in view of the paramount importance of these inputs in business economics. List of references Baumol, W.J., S.A.B. Blackman and E.N. Wolff, 1989, Productivity and American Leadership. (MIT Press, Cambridge, Massachusetts). Bovenberg, L., and L.H. Goulder, 1991, “Introducing intertemporal and open economy features in applied general equilibrium models”,De Economist, Volume 139, Number 2. Ees, H. van, H. Garretsen, L. de Haan and E. Sterken, 1996, Investment and Debt Constraints; evidence from Dutch panel data, mimeo, Groningen University. Englander, A.S., and A. Mittelstädt, 1988, “Total Factor Productivity: Macroeconomic and Structural Aspects of the Slowdown”. OECD Economic Studies, Number 10, Spring. Gertler, M., 1988, “Financial structure and aggregate economic activity”, Journal of Money, Credit and Banking, Volume 20, pp. 559-588. Harper, M.J., 1997, Estimating Capital Inputs for Productivity Measurement: An Overview of Concepts and Methods. Paper prepared for the Conference on Measuring Capital Stock, Canberra, March 10-14, 1997. Hasan, M.A., and S.F. Mahmud, 1993, “Is Money an Omitted Variable in the Production Function? Some Further Results”. Empirical Economics, Volume 18, Number 3. Hulten, C.R., 1996, “Capital and Wealth in the Revised SNA”. In: J.W. Kendrick (ed.), The New System of National Accounts. (Kluwer Academic Publishers, Boston/Dordrecht/London). Jorgenson, D.W., 1990, “Productivity and Economic Growth”. In: E.R. Berndt and J.E. Triplett (eds.), Fifty Years of Economic Measurement. (University of Chicago Press, Chicago). Keuning, S.J., “Productivity Changes and Shifts in the Income Distribution; with an application to the case of Indonesia”. Economic Systems Research, Volume 7, Number 3. Keuning, S.J., 1996, Accounting for Economic Development and Social Change, IOS Press, Amsterdam etc. Keuning, S.J., and T. Reininga, 1997, Accounting for the Use of Financial Capital as an Input in Production, National Accounts Occasional Paper Series No. NA-085. (Statistics Netherlands, Voorburg/Heerlen). Maddison, A., 1987, “Growth and Slowdown in Advanced Capitalist Economies: Techniques of Quantitative Assessment”. Journal of Economic Literature, Volume XXV, Number 2. Miller, M.H., and F. Modigliani, 1966, “Some Estimates of the Cost of Capital to the Electric Utility Industry, 1954-57”. American Economic Review, Volume 56, Number 3. Rymes, T.K., 1983, “More on the Measurement of Total Factor Productivity”. The Review of Income and Wealth, Series 29, Number 3. Scott, M.FG., 1993, “Explaining Economic Growth”. American Economic Review, Volume 83, Number 2. Stiglitz, J.E., 1992, “Capital Markets and Economic Fluctuations in Capitalist Economies”. European Economic Review, Volume 36. Triplett, J.E., 1996, “Depreciation in Production Analysis and in Income and Wealth Accounts: Resolution of an Old Debate”. Economic Inquiry, Volume 34, January. 12 United Nations, Eurostat, International Monetary Fund, Organisation for Economic Co-operation and Development, and World Bank, 1993, System of National Accounts 1993. Series F, No. 2, Rev. 4. (United Nations, New York, etc.). Yeager, L.B., 1979, “Capital Paradoxes and the Concept of Waiting”. In: M.J. Rizzo (ed.), Time, Uncertainty and Disequilibrium (D.C. Heath, Lexington). 13 View publication stats