«Jonathan Haskel Imperial College Business School, CEPR, Ceriba, IZA and UK-IRC Keywords: copyright, IP, innovation, knowledge, investment, ...»
We now have a new dataset for market sector investment in artistic originals from 1990. In the following section we use our new central estimates of GFCF to construct a capital stock for artistic originals in the UK market sector, where the market sector is defined as sections A-K & OP according to SIC2003. Since the starting values for the capital stock require a longer back-series for GFCF, we backcast the series using growth rates of GFCF in originals from the National Accounts. In the case of film, we extend the series using the data on joint US/UK production presented earlier. 48 Of course this assumes that UK investment grew at a similar rate to UK investment in US/UK co-productions, which may not be the case. Finally we deflate our investment data and construct estimates of the capital stock and compare them with those implied by official data recorded in the National Accounts.
15.1. Deflation Since we need to construct real measures of the capital stock, we first need to deflate our investment data to estimate real GFCF in originals. For some assets, transactions data are available, for instance, commissioned or acquired broadcasting originals. However, since each original is by definition a unique asset, separating price and volume is challenging, particularly if the quality of originals has changed over time. Before reviewing what options are available, it is worth returning to the framework to confirm exactly what we wish to measure.
The first point to make is that we know the equilibrium relationship between downstream
revenues and upstream creation can be summarised as in equation (2) and repeated below:
P(N)N=ΣP(R)R/(1+r)t This states that the asset value of the good must equal the discounted rental payments from the users of the good. However, it is important that data on P(R)R, and therefore P(R), are reflective of true rentals. That is, payments for rights to use for less than one year with no transfer of economic ownership. Payments for ownership and/or payments for multiple years represent a component of P(N)N, from which a rental must be backed out. When estimating prices it is important not to confuse P(R)R with P(N)N or some hybrid of the two. We need to be aware that the price paid for the use of originals (the rental price) is not the same as the price of the asset. The relationship between the asset price and the rental price is given by the
user cost relation:
Although this data go back as far as 1915, our series for the dollar/sterling exchange rate only goes back as far as 1945, and our value-added deflator as far as 1971. Therefore we use 1971 as our initial value.
P(R) = P(N) (ρN+δR) Where ρN is the real rate of return in the N (upstream) sector and δR the rate of deprecation for the relevant type of capital or artistic original.
To derive a measure of real GFCF we need to measure P(N), the asset price. If this is not available we can instead estimate the rental price and transform that into an asset price using the user cost relationship defined above.The issue we face is that little data on asset transactions exist, since much of it is produced on own-account and used in the same organisation. Some limited information on prices for use are available, but that is not the same as the price of investment. Ideally we need the latter to convert our data for nominal GFCF to real volume measures. So what options are available to us?
One is to use the implied investment price indices in the ONS data on investment in artistic originals. However, we know little on how those data are constructed and whether they accurately reflect P(N). A second option would have been to use estimates implied by revenue schedules in data from the Collecting Societies, but this option is not available to us for now. This would give estimates of P(R), which could be converted to P(N) with the user costs relation.
A third would be to develop an input costs deflator for specific assets. This is common in areas where little data exist on market transactions, such as own-account software investment.
There are two possibilities. One would be to weight together indices for various upstream inputs including labour, capital and materials, where the weights are based on upstream expenditure shares. As second would be to develop an index using just the wages of occupations that work in asset creation, or use a more general measure based on the Average Earnings Index (AEI). Since around 55-65% of value-added flows to labour (depending on how capital is defined) this basically assumes that upstream asset prices are driven by the price of upstream labour. In theory the wage index should also include some estimate or assumption for labour productivity. In practice, this would mean a mismeasurement in price changes equal to the error in the productivity assumption, plus an error from the missing price contributions of other factor and intermediate inputs.
A fourth option would be to use the deflators developed by the BEA in their recent developmental work (Soloveichik, 2010), where the assumption would be that the marketplace for artistic originals is open and internationally competitive, which would appear reasonable. The type of price index used by Soloveichik varies by asset. In the case of music, it is based on the earnings of musicians, as in the third option described above.
A fifth option is to use an implied GDP deflator, and assume that general prices also reflect the price of originals (Haskel et al, 2009). Finally a sixth option is to use consumer price indices for re-productions or copies of the relevant asset types.
The OECD recommend that in cases where GFCF is measured as the NPV of royalty payments, then the data on rentals can be decomposed into price and volume. For estimates based on alternative methods, such as upstream input costs, the OECD suggest either the use of consumer price indices for copies of originals, or a measure based on the input prices for factors and goods used in the production of originals.
To say a little more about the sixth option, a common misconception is that rental prices are observed in the sale of copies of originals. That is, that rental prices can be observed in prices of cinema tickets, books, CDs, etc. But as the framework makes clear this is P(Y), not P(R).
Now it may be that P(R) makes an important contribution to P(Y), but so will the prices paid to labour, physical capital and other forms of knowledge capital used in the downstream sector. Additionally P(Y) will be affected by total factor productivity in the downstream sector. 49 In some cases it will be possible to make use of some information or assumption on the industry or asset in question. For instance, in the film industry, the rental fee paid by the
cinemas is usually around 30% of revenues from ticket sales (Soloveichik, 2010). So:
P(R)R = λ.P(Y)Y
Where λ is some constant proportion. But note that the relationship is between revenues, not prices. To calculate the relationship between prices we need some information on R, that is, the real stock of artistic originals, for which we need a price index to calculate. So there is a problem of circularity. Also note here that P(Y)Y refers to revenues from cinema ticket sales, not cinemas, since the latter would include revenues from food and all other goods and services provided. Also note that the true downstream is not just cinemas and instead includes DVDs, TV licensing etc.
So it is not conceptually correct to assume that the price of copies changes at the same rate as the price of the underlying asset, since the price of copies is determined by the price of factor Likewise P(R) is partly determined by TFP in the upstream sector. For further information see Corrado, Goodridge and Haskel (2010).
or intermediate inputs and technological progress in the downstream, whilst the price of assets is determined by the same factors but in the upstream. That is, the evolution of the price of say CDs reflects innovation and productivity in the manufacture and publishing of CDs, rather than productivity in the creation of the underlying knowledge (the music) itself.
As an aside, potential exceptions to this include services such as Netflix, which provides customers with downloads of movies for a fee. The difference is that in the case of Netflix, there are far less factor and intermediate costs than in other parts of the downstream. For instance, they do not require buildings to project films, or to display DVDs for sale or rental.
Therefore, since we would expect P(X)XY to be much smaller, P(R)R will be a much larger component of P(Y)Y. So although conceptually it would still be incorrect to use the price of Netflix download, in a practical sense it could be used as a proxy if it is considered that output prices are largely driven by rentals for film originals. The issue with using data for Netflix, or similar services for other assets, is that they are based on new technologies and capabilities, so the length of any time-series would be extremely limited.
Figure 36 shows retail price indices for i) Copies of Music Originals: CDs etc. ii) Copies of Books and Newspapers and iii) BBC Licence Fee, and emphasises the conceptual point made above. Compare the RPI for Books and Music. As can be seen, the price index for copies of Books, an ‘old technology’ has been growing consistently throughout the period presented.
In contrast, the price index for music formats grew much more slowly and started to decline from 1998, reflecting improved technologies in the downstream production of copies, falling prices in ICT capital and materials, and possibly falling prices for rental of originals. Only the latter technical progress in the creation of music or literary works, the rest relate only to downstream activity.
Figure 30: ONS Retail Price Indices (1990=1) 2.5 0.5 Source: ONS Monthly Digest of Statistics Note to figure: Retail Price Indices for ‘CDs, records and tapes’ (DODA), ‘Books and newspapers (DODC), ‘Television licences’ (DODE). Each index has been re-based so 1990=1.
As we keep emphasising that downstream rental prices reflect upstream asset prices, which in turn reflect input prices and technical progress in the upstream, it is worth highlighting what we think we already know about the contributions of these. Corrado, Goodridge and Haskel (2010) show the price of knowledge investments to have fallen in the context of UK R&D, with the hypothesis being that the emergence of the internet and other communications technologies has made the conduct of upstream research much cheaper. It seems reasonable to suggest that such technologies have had a similar impact in the production of artistic originals, for example in the case of books/scripts, where surely the upstream productivity of formulating ideas, developing those ideas through research, communication with up and downstream participants, and collaboration has improved. Speculatively, it might be expected that the price of artistic originals has risen less fast than the price of all goods and services in general.
Another option considered was to use the deflators currently used for artistic originals in the UK National Accounts. Figure 37 shows three deflators implied by current and constant price ONS GFCF data for artistic originals, alongside the value-added deflator used for artistic originals and other knowledge assets in constructing the NESTA Innovation Index. As can be seen, official data suggests prices for artistic originals rising significantly faster than goods and services in the wider economy. This goes against the view that the price of knowledge is rising less fast than those for final goods, or even declining, due to the revolutions that have occurred in the communication and sharing of knowledge since the emergence of the internet.
(Corrado, Goodridge and Haskel, 2010).
Figure 31: Deflators for Copyright GFCF 2.4
Note to figure: The three deflators presented here are implied deflators for Publishing, Broadcasting and Recording, derived from the CP and KP GFCF data. The very small differences are likely due to rounding, suggesting the same deflator is used for Film, TV & Radio, Literary and Recording Originals The NESTA series above is an implied deflator for aggregate GVA for a specific definition of the market sector (A-K excl dwellings), as used in the Innovation Index (Haskel et al, 2009).
We have argued that a deflator based on the price of copies does not reflect the price of the asset, and that instead it reflects input prices and productivity in the production of the copy itself. We also consider the ONS deflator to be rising too fast in light of our previous work.
Therefore, as in previous work, we use an implied price index for aggregate value-added as our preferred deflator.
15.2. Depreciation In order to construct the capital stock we need some estimates of depreciation rates, ideally by asset type. Following Corrado, Hulten and Sichel (2006) the rate used for artistic originals in the NESTA project and other UK work on Intangible Assets was 20%. The rate used in the ONS Volume Index of Capital Services (VICS) is 13.33% (Long et al, 2010).
Returning to the recommendations of the Eurostat Taskforce, their report suggested that depreciation rates ought to be fast in the early years, and that life-lengths should be set at
between 5-10 years. Two possible functions mentioned were:
− Linear rate, with suitable Winfrey retirement function − Geometric rate, with at least a double-declining balance Note that a life-length of just five to ten years implies a very high geometric rate of depreciation. Such assumptions on lifespan in part go against work by the BEA, discussed below, as well as copyright law itself, since the latter typically assigns property rights of 70 years after death to authors, and 50 years after performance to performers.
It could be argued that since many artistic originals are failures or 'flops', then a high depreciation rate should be used to reflect this. Instead it is more helpful to consider the rate as a weighted function, where the weights depend on sales or rentals of copies or versions.