Technological Change and the Stock Market: A Comment Stephen Wright¤ Department of Economics Birkbeck College, University of London London WC1 E7HX e-mail:
[email protected]; May 14, 2004
Abstract In a recent paper Laitner & Stolyarov (2003) assert that measured Tobin’s q has usually been well above 1, and use this as supporting evidence for their empirical …nding that there are signi…cant quantities of unrecorded intangible assets. This key feature of q turns out however to be due primarily to the omission of important elements of tangible assets from the denominator. The corrected q series turns out to have a mean well below unity. The statistical basis for their other empirical results suggesting sign…cant quantities of intangible capital also appears fragile. ¤
I am grateful to John Laitner and Dmitry Stolyarov for their openness in discussing their original work and for providing me with their F ortran program; to Michael Palumbo of the Federal Reserve Board’s Flow of Funds Section for his very friendly and helpful clari…cation of some data-related queries; and to Paul Schweinzer for programming assistance.
1. Introduction In a recent paper in this journal Laitner & Stolyarov (2003) one of the authors’ principal conclusions is that there is a signi…cant amount of unrecorded intangible capital. They draw this conclusion from two pieces of evidence. The …rst is that, using their dataset, measured Tobin’s q has usually been well above 1. This key feature is demonstrated by a chart (their Figure 1) that is referred to repeatedly in the paper to motivate their analysis. 1 In addition to the indirect evidence from q, the authors derive quantitative estimates of the magnitude of intangible assets from method of moments estimates of key parameters in a growth model. The resulting estimates, the authors claim, have the advantage (unlike the q data) of being independent of measured capital stock data. On closer examination, both pieces of evidence prove distinctly fragile, and for closely related reasons. The feature that their measure of q (which they calculate for the business sector as a whole) is typically above unity turns out to be entirely dependent upon a number of clear errors in the way that the authors construct their data. They both overestimate the numerator and underestimate the denominator of q. The latter error is most signi…cant: the primary factor being the omission of signi…cant elements of tangible, rather than intangible assets - most notably land. When the calculation is carried out correcting for these errors the resulting q series turns out to have a mean well below unity. This feature of the data is even more marked when q is calculated using published data from the Federal Reserve balance sheets just for the non-farm, non-…nancial corporate sector.2 The authors’ quantitative estimates of the share of intangible capital in total capital are indeed not directly derived from measured capital stock data; but it is straightforward to show that they are they not entirely independent either. Just as important, they are subject to an important degree of re-interpretation: namely, as the share of intangible capital plus land. On the basis of available estimates of 1
To quote from their abstract; from their opening paragraph; and from Section IV (page 1258) The stock market value in the numerator of q re‡ects ownership of physical capital and know ledge, but the denominator measures just physical capital. Therefore q is usually above 1....This paper attempts to provide a simple yet comprehensive model which we can use to explain and interpret the average and time series features of the data....Figure 1 suggests .. that the stock of applied know ledge ( A) is 30-50 percent as large as the physical capital stock ( K)... 2 This feature is discussed at some length in Smithers & Wright (2000) (cited by the authors), and is also well known to Fed statisticians.
2
the share of land in business tangible assets, their own estimates leave little room for signi…cant quantities of intangibles. The main focus of this note will be on the problems with the authors’ q data. In Section 2, the elements of the comparison between the original and corrected series are presented in a sequence of charts. The details are in general somewhat tedious and are relegated to an appendix. Section 3 comments brie‡y on the quantitative estimates of the share of intangibles. Section 4 concludes.
2. Comparing q Estimates 2.1. Replicating Laitner & Stolyarov Tobin's q for the Business Sector, as in Laitner & Stolyarov (2003) 2.500
2.000
1.500
1.000
0.500
0.000 195204
195704
196204
196704
197204
197704
Laitner & Stolyarov
198204
198704
199204
199704
200204
Recalculated Laitner & Stolyarov
Figure 2.1: Figure 2.1 shows Laitner & Stolyarov’s original q estimate. The chart also shows an updated estimate on the latest data.3 The correspondence between the original and recalculated series over their common data sample is extremely close. 3
Note that the recalculation does not follow the formula as reported in their original paper,
3
Both clearly share the feature they allude to that most observations are greater than unity. This feature is, unsurprisingly, particularly marked in recent years. 2.2. Correcting the numerator The numerator of this measure of q is intended to be an estimate of the market value of the entire business sector, made up of both corporate and unincorporated businesses. Laitner & Stolyarov calculate this value by residual from the net wealth of the personal sector, and the net liabilities of the government, overseas and the monetary authority, taken from the Flow of Funds tables. This apparently simple formula has the following drawbacks: 1. It includes the market value of holdings of equities of overseas corporations by US residents as well as the (empirically trivial) value of gold and SDRs; 2. It includes the value of net overseas direct investment by US corporations, and is therefore not directly comparable with the capital stock data; 3. It imputes all unidenti…ed …nancial liabilities to the business sector. The …rst two of these are clear errors; the third is at best a contentious assumption. An alternative estimate of the numerator that does not have these drawbacks can be constructed relatively straightforwardly (if more laboriously) by adding up individual components using data from the Flow of Funds tables (for details see Appendix A, which also shows the identities linking the original and corrected market value series). 2.3. Correcting the denominator The rationale of q implies that the denominator should capture all the physical assets owned by the domestic business sector. The denominator used by Laitner & Stolyarov is given by the sum of the private nonresidential …xed capital stock and nonfarm inventories. This measure too has very clear-cut drawbacks: 1. It contains some nonresidential …xed assets that should not be included, namely those belonging to households and non-pro…t making institutions. which also contains a number of errors - see Appendix A. Laitner & Stolyarov have con…rmed that the approach used here replicates the actual, rather than reported formula used in their data construction.
4
2. It omits residential capital (dominated by tenant-occupied housing owned by the noncorporate business sector) 3. It omits the value of land. Most of the missing elements turn out to be readily available in published data (both in the Federal Reserve Flow of Funds tables, and in BEA data). Where there are no data, informed guesswork can be applied that follows as closely as possible the methodology used by the Fed in constructing their existing balance sheets for the non-farm, non-…nancial corporate and non-corporate sectors. Since these two sectors own nearly 90% of total estimated business tangible assets, the results are not sensitive to the inclusion of those missing elements for which there are no available data. If anything there is some reason to think that the guessed-at elements of the denominator are probably underestimated 2.4. The impact of the corrections
Tobin's q for the Business Sector: Impact of Corrections to Laitner & Stolyarov's data 2.500
2.000
1.500
1.000
0.500
0.000 195204
195704
196204
196704
197204
197704
Recalculated Laitner & Stolyarov
198204
198704
Corrected Numerator
Corrected Numerator and Denominator
Figure 2.2: 5
199204
199704
Figure 2.2 shows the impact of the two sets of corrections alongside the recalculated estimate using Laitner & Stolyarov’s approach. The second line shows the impact of amending the formula for the numerator, which has a fairly modest impact except in recent years. The third line shows the much more signi…cant impact of correctly de…ning the denominator - with the inclusion of residential capital and land having more or less equal impact. The net result of all the corrections is that at the peak of the market in 2000 Laitner & Stolyarov’s q estimate was around 60% higher than the corrected measure using identi…ed market value. The chart also clearly shows that the corrected q series, far from being predominantly above unity, is predominantly below. Figure B.1 in the appendix shows the implied sectoral q …gures that underlie this aggregate …gure. The appendix also provides a summary table of descriptive statistics.
3. Method of Moments Estimates of the Share of Intangibles Laitner & Stolyarov’s empirical results in Sections III and IV do not depend directly on recorded capital stock data - a feature of their results they claim is an advantage (see their footnote 22). Their results do however depend on recorded capital consumption data that are calculated using the same methodology as recorded capital stock data. Given this, it is quite easy to show that there is a strong link between their empirical results and the failings of their q data. While the authors’ method of moments estimation procedure estimates six parameters simultaneously, applying cross-equation restrictions across the six equations of their model, there is one key empirical relationship that determines ® A µ ´ ®+¯ ´ A+K ¤ ; the share of the intangible capital stock in total capital. Working in terms of steady state ratios for simplicity, substituting from the authors’ equation (19) into (22) implies £ ¤ IK = (1 ¡ µ) g M f+± M
(3.1)
where (using their de…nitions), I K is gross recorded investment, M is market value, ft¡1 Mt¡M gM is a “revolution-adjusted” measure of growth of market value, that f= f Mt¡1 strips out the one-o¤ e¤ect of the single technological innovation;4 and ± is the 4
gM f > gM ´
Mt ¡Mt¡1 ; Mt¡1
actual market value growth.
6
true depreciation factor. The estimate of µ is greater than zero because recorded investment is “too low” in relation to market value to be consistent with growth rates and depreciation. The authors do not include K ¤ in their empirical model directly, but do include it indirectly, since they have an equation for recorded depreciation, given in their model by D = ±K ¤ where ± > ± is the average depreciation rate. (In their framework this is higher than true depreciation due to the impact of periodic technological innovations). Using this de…nition, the evolution of measured tangible capital (excluding land) implies IK = gK ¤ + ±: K¤
(3.2)
Combining (3.1) and (3.2) implies 1¡µ =
µ
1 q LS
g K¤ + ± gM f+±
¶
(3.3)
where q LS = M=K ¤ is measured Tobin’s q on the authors’ de…nition. The second ratio on the right-hand side of (3.3) is very close to unity; 5 indeed it should be precisely unity if M and K ¤ are not to drift apart inde…nitely. In Laitner & Stolyarov’s framework M typically grows more rapidly than K ¤; and depreciates more slowly; but is lowered periodically by structural shocks when “revolutions” happen; the higher depreciation rate for K ¤ re‡ects the average e¤ect of these shocks, since the level of K ¤ is never hit by these negative shocks. Thus, on a single equation basis, the fact that µ is estimated at less than unity relates directly to qLS being above unity on averag (in practice cross-equation restrictions complicate matters somewhat). Given the explanation for the authors’ q …gures being above unity as set out in Section 2, one obvious explanation of the estiamate of µ being greater than zero may well be that it re‡ects, wholly or in part, “missing” tangible as well as intangible assets, most obviously the exclusion of land. Figure 3.1 shows that, using available data, the share of land in total tangible assets appears to lie in a range 10-20%.. This …gure is largely driven by the Fed’s estimate of the quite large share of land in the capital of the non-corporate business sector (for details, and a chart of sectoral …gures, see Appendix). If anything there 5
My best guess at a point estimate, using the authors’ empirical estimates and appropriately adjusted growth rates, is 1.004.
7
Estimated Share of Land in Business Tangible Assets 25.0
20.0
15.0
10.0
5.0
0.0 195204
195704
196204
196704
197204
197704
198204
198704
199204
199704
200204
Figure 3.1: is a suspicion, from looking at the underlying sectoral pattern, that this …gure is on the conservative side.6 Using Laitner & Stolyarov’s own results, it is possible to derive a perhaps rather crude, but nonetheless revealing test of the hypothesis that land is the sole explanation for the estimated value of µ lying above zero. With three forms of capital, the steady state value of µ will be given by µ=
A+L A + L + K¤
where A is intangible capital and L is land. Since we have data on L=(L + K ¤ ); which using the data in Figure 3.1 has a sample average of 14%, the hypothesis µ that A = 0 implies µ = :14; hence we have H0 : ® = 1¡µ ¯ = :167¯; a simple linear restriction. The authors have kindly provided an amended set of empirical results (and the Fortran program that generates these) in which investment and 6
In particular the Fed’s data show non…nancial corporations as owning virtually no land at all in recent years.
8
depreciation …gures are adjusted from the original series to include investment in residential capital, consistent with market value data. The change in the resulting parameter values is small compared to the published values; but the mutual consistency of the data series used implies that the resulting estimate of ® can be interpreted directly as the exponent on intangibles plus land. Their point estimate ® of µ = ®+¯ is 0.31; but using parameter variances and covariances to construct b results in a t-statistic of the standard error of the linear combination ® b ¡ :167¯; just 1.84 for the test of the implicit null that A = 0: This seems a rather slender statistical thread on which to hang an assertion that there is a signi…cant quantity of unmeasured intangible capital.7
4. Conclusions This note has drawn two key conclusions that cast signi…cant doubt on Laitner & Stolyarov’s assertion that there are is evidence of a signi…cant quantity of intangible assets. First, Tobin’s q for the business sector, once correctly calculated using available data, does not typically lie above unity, but instead has normally been somewhat below. Second, even when the authors estimate the share of intangible assets by a more indirect method, from method of moments estimates of the parameters of a growth model, their estimate must be reinterpreted as the share of intangibles plus land; and the statistical basis for concluding that there is a positive residual element of intangibles appears distinctly fragile. This does not, of course, mean that intangible assets do not exist; it just means that if they do there must be something wrong with the published data. Two possible explanation for this might be that the BEA’s capital stock data may be based on depreciation rates that do not re‡ect true economic depreciation, or that investment de‡ators may be incorrectly calculated (as proposed, for example by Gordon, 1990). 7
Given the cross-equation restrictions, there are other candidate explanations for a positive estimate of µ; and hence ®. In particular, the authors impose constant returns to scale, implying that the labour share is constrained to equal 1 ¡ ® ¡ ¯; which may also tend to push the estimate of ® away from zero if, as is commonly found, freely estimated Cobb-Douglas parameters on labour and tangible capital alone would imply decreasing returns.
9
APPENDIX A. Data Construction Table A1 (appended to paper) sets out the sources and data construction methods, which allows a comparison of the q estimates used in this paper with those in Laitner & Stolyarov (2003) (hereafter LS). The panels of Table A1 correspond to equivalent sections of a spreadsheet that can be downloaded from www.econ.bbk.ac.uk/faculty/wright. All source references therein are to Z1 (‡ow of funds) tables (Federal Reserve, 2004), except those to TA, which refer to the BEA tangible assets tables, and NIPA, which refer to National Income and Product Accounts tables. A.1. Market Value A.1.1. Replicating Laitner & Stolyarov Panel A of Table A1 reproduces verbatim the formula for market value using the Flow of Funds table references as given in LS (P1261, footnote to Table A1). However, this formula results in a series which appears radically di¤erent from their own reported series for market value (in their Appendix A, col. 2), with the series constructed using their reported formula exceeding their reported series by anything up to 50%. The authors have con…rmed, however, that their reported formula, as published, contains a number of errors. Panel B provides an alternative, simpler, formula, deriving the value of the business sector by residual from the net worth of the household sector and nonpro…t making institutions, less the net liabilities of the government, monetary authorities and overseas. This results in a series very much closer to that reported in LS in recent years (di¤ering by only fractions of a percent for the past decade).The authors have con…rmed in email correspondence that the approach applied in Panel B is that actually applied in their dataset. A.1.2. Market Value from Identi…ed Components Panel C provides de…nitions underlying an alternative market value series that corrects the problems with the LS approach. It builds up the series by adding up identi…ed components of market value using ‡ow of funds data on equities, assets
10
and liabilities.8 These sum to a series which measures the identi…ed value of USowned business. To ensure complete comparability with tangible assets data, the data are then adjusted to provide the market value of domestic US business (ie, including the value of foreign companies’ operations in the US, but excluding the value of US companies’ operations abroad). Although this last adjustment is required in logic, it turns out to make (surprisingly) little di¤erence to market value estimates. Panel D shows how this approach can be reconciled with that of Laitner & Stolyarov. Their series does not correct for net ODI, and includes both US holdings of overseas equities and the very small category of other assets (gold and SDRs) not included in liabilities data. Their estimate also includes unidenti…ed liabilities: the gap between total recorded assets and the sum of recorded liabilities and all assets not included in liabilities. This unidenti…ed element in debt may indeed re‡ect unrecorded liabilities of US business, and thus it is of interest to see its impact; but anyone using this approach should be aware of its inclusion. The …gures show that on average inclusion of unidenti…ed debt slightly lowers measured q (since for most of the sample unidenti…ed debt was negative); however, this element switched sign quite dramatically in the 1990s, and by end-2002, unidenti…ed debt was equivalent to 14% of identi…ed business market value. A.2. Reproducible Capital at Replacement Cost Panel F of Table A1 reproduces the formula for business …xed capital plus inventories used in LS: the resulting series is virtually identical . Panel G provides an alternative formula based on BEA data. There are two key modi…cations.First, that part of nonresidential …xed capital belonging to the personal sector and non-pro…t making institutions is excluded, for consistency with the derivation of market value data. Second, a series is constructed for the residential capital stock of the business sector. Following the same methodology as outlined by the Federal Reserve in constructing their tangible assets series (Federal Reserve, 2000, p299) for the nonfarm, noncorporate business sector, this subtracts the residential capital stock of non-pro…t making institutions, and of owner-occupied housing, from the private sector total (unlike the Fed approach 8
The …nancial sector’s recorded assets and liabilities are measured, for convenience, by residual from total identi…ed stocks and the equivalent stocks of other sectors. As such it simply inverts the identities used to derive the aggregates, and does not impute any residual items to the …nancial sector. Note that data for recorded …nancial assets are corrected for an error (con…rmed by the Fed) in the published version of Table L5 (Jan 2004 release).
11
tenant-occupied farm housing is retained since the de…nition here includes all farms). Total private inventories are then added to the …gure for …xed capital. The resulting series for total business reproducible capital is systematically larger than the series used by LS, di¤ering by a factor of around 20% in the early part of the sample, with the di¤erence falling steadily to around 10% by the end of the sample. It is extremely close to the equivalent …gure derived on a Z1-equivalent basis, as described in the next section, never di¤ering by more than around 1%. A.3. Tangible Assets Including Land
Implied and Imputed Shares of Land in Tangible Assets 45.0 40.0 35.0 30.0 25.0 20.0 15.0 10.0 5.0 0.0 -5.0 -10.0 195204
195704
196204
196704
197204
197704
198204
198704
199204
199704
nonfarm noncorporate business (Z1)
nonfarm nonfinancial corporate (Z1)
financial corporations (imputed)
farms (imputed)
200204
farms (imputed using mortgage data)
Figure A.1: Panel H of Table A1 summarises the construction of a series for business tangible assets including land. The series is built up from two elements (for the 12
non-farm, non-…nancial corporate and non-corporate sectors) that can be taken direct from the ‡ow of funds tables, plus equivalent series for …nancial corporations and farms, the construction of which (involving some guesswork) is summarised in Panels M and N. Figure A.1 shows that the implied land series for the two sectors for which the Fed produces tangible assets data both show a distinct discontinuity after 1989, following a change in Fed methodology. 9 The implied share of land for non…nancial corporates falls virtually to zero in the mid-1990s, but then recovers somewhat. The implied share of land for non-corporates shows a similar fall, but then recovers distinctly more strongly, to a much higher level (re‡ecting the much more signi…cant proportion of residential capital for the non-corporate sector). Panel M details the (fairly straightforward) construction of estimates of tangible assets for …nancial corporations. Data for structures and equipment and software come straight from the BEA nonresidential tangible assets series (Table 4.1). Both inventories and residential capital are set to zero, since these are already fully allocated in the ‡ow of funds tables. Real estate …gures are estimated by scaling BEA structures data using the ratio of real estate to structures for non…nancial corporations. As Figure A.1 shows, the resulting imputed share of land in tangible asesets is very similar to that for non…nancial corporations (the somewhat higher …gure in earlier years re‡ecting the lesser importance of equipment and software of …nancial corporations compared to non-…nancials). Panel N details the equivalent calculation for farms (both corporate and noncorporate). BEA tangible assets …gures and inventory are available for all categories, except for the very small category of residential plant and equipment (durable goods in rented properties) which is imputed assuming the same ratio as for non-farm non-corporates.. The implied real estate …gure is calculated using BEA structures …gures on the same basis. Figure A.1 shows that the resulting …gure for the value of farm land appears distinctly conservative, since it is well below the share of land for non-farm non-corporates, despite the clearly land-intensive nature of farming. As an alternative method of calculation, to provide at least a basis for comparison, farm real estate values can be inferred on the basis of farm and household sector mortgages (lines 5 and 2 of Z1 Table L217, respectively), by working on the assumption that collateral ratios applied by lending institutions are the same, using the ratio of home mortgages to household real estate (Table L100, line 4). 9 See Wright (2004) for a discussion and a proposed alternative treatment that deals with land more consistently.
13
Land values can then inferred as the di¤erence between the resulting real estate values and BEA data on farm structures. Figure A.1 shows that this results in distinctly higher implied land values in the early part of the sample, but that the implied share of land then falls steadily, such that, by the end of the sample, it is slightly negative - which would imply that the BEA’s estimate of the replacement value of farm structures is greater than their market value including the land they sit upon.10 However, Figure A.1 in Appendix B shows that even this very extreme, and distinctly pessimistic implication for farms has very limited implications for the resulting q estimate for the non-corporate plus farm sector in aggregate.
B. Alternative q Estimates Implied Sectoral q Values 1.8
3.0
1.6
2.5
1.4
2.0
1.2
1.5
1.0
1.0
0.8
0.5
0.6
0.0
0.4
-0.5
0.2
-1.0
0.0 195204
-1.5 195704
196204
196704
197204
197704
198204
198704
199204
199704
200204
nonfarm nonfinancial corporations noncorporate business + farms noncorporate business + farms (mortgage-based estimate of farm real estate) financial corporations (rhs)
Figure B.1: 10
This somewhat surprising result does not appear to be due to a fall in the importance of farm mortgages as a form of borrowing - on the contrary, the share of mortgages in total farm liabililities was if anything on a somewhat upward trend through the data sample.
14
An advantage of building up both numerator and denominator from their sectoral components is that it is possible to examine the implied sectoral q estimates that underlie the aggregate …gures. These are shown in Figure B.1. The …gure for the non…nancial corporate sector is, as already noted, well below unity for most of the sample.11 The impact of this on the aggregate measure is somewhat o¤set by the implied q …gure for the noncorporate sector (including all farms) which is very stable, and uniformly somewhat above unity;12 and, on a relatively few occasions, by the much more volatile (and hence almost certainly less reliable) …gure for the …nancial sector. Table B1 summarises the key features of the various alternative q estimates. The table makes clear in summary form that, once q is measured correctly, the key feature of Laitner & Stolyarov’s dataset, of a mean greater than unity for the business sector as a whole disappears completely. Instead the corrected series has a sample mean well below unity. The table shows that this is largely driven by the changes to the denominator. Table B1. Alternative Estimates of Tobin’s q
¤
Geometric Averages 1953-199513 1953-2000 All business, Laitner & Stolyarov (2003) 1.105 1.172 All business, Laitner & Stolaryov updated 1.134 1.200 All business, Corrected Numerator 1.110 1.161 All business, Corrected Numerator and Denominator 0.816 0.859 Ditto, including Unidenti…ed Liabilities in Numerator 0.785 0.834 Nonfarm, Non…nancial Corporations 0.614 0.670 Noncorporate business + farms 1.158 1.149 Ditto, using alternative estimate of farm real estate 1.128 1.126 ¤ Financial Corporations 0.241 0.411¤ Arithmetic averages
Table B1 also shows that, within the sectoral q estimates, the feature of an 11
Smithers & Wright (2000) and Wright (2004) discuss possible explanations for this feature, which Wright (2004) shows is also clearly evident in data for quoted companies over a much longer sample. 12 The chart also shows the very limited impact on noncorporate q of using the alternative estimate of farm real estate values, as discussed in Appendix A. 13 This truncated sample is used for most of Laitner & Stolyarov’s paper, since they believe later data may have been a¤ected by a bubble.
15
above-unit mean q only appears to survive for the noncorporate sector - not generally regarded as the most likely sector to have signi…cant intangible assets. But, since the numerator of this estimate is not directly based on market valuations it is in any case very hard to interpret; it seems more likely that the above-unit mean results from a conservative approach to the valuation of non-corporate land. The …nancial sector appears, like the non…nancial corporate sector, to have a q signi…cantly below unity on average, but, due to its volatility (including some negative observations early in the sample) this series is likely to be less reliable.
References [1] Federal Reserve (2000) Guide to the Flow of Funds Accounts, Board of Governors of the Federal Reserve System, Washington DC [2] Federal Reserve (2004) Flow of Funds for the United States January 2004, Board of Governors of the Federal Reserve System, Washington DC [3] Gordon, R J (1990) “The measurement of durable goods prices” National Bureau of Economic Research Monograph series Chicago and London: University of Chicago Press. [4] Hall, R (2001) “The Stock Market and Capital Accumulation,” American Economic Review 91:1185-1202 [5] Laitner, J and Stolyarov, D (2003), “Technological Change and the Stock Market”, American Economic Review, vol. 93, no. 4 1240-67 [6] Smithers A, and Wright S (2000) Valuing Wall Street McGraw-Hill, New York [7] Wright, S (2004) “Measures of Stock Market Value and Returns for the US Non…nancial Corporate Sector, 1900-2002” Birkbeck College, University of London (2nd stage revision for Review of Income and Wealth completed and resubmitted).
16
Table A1 Data Definitions for Comparison with Laitner & Stolyarov (2003) A.Market Value as defined in Laitner & Stolyarov less Less less Source: L100, R1 L100, L106, L105, R25 R15 R18
plus L105,R7
plus L105,R1 0
Plus L108, R10
minus L106, R14
Minus L108, R15
plus L107, R1
minus L107, R23
Equals
Househol ds and non profit
Househol ds and non profit
Fed governm ent
State & local governm ent
State & local governm ent
State & local governm ent
Monetary Authority (excl FRB)
Fed governm ent
Monetary Authority (excl FRB)
Rest of World
Rest of World
Total Business, L&S definition
Financial Assets
total liabilities
Treasury currency (liabilitie s)
Credit market liabilities
US gov't securities (assets)
municipa l securities (assets)
US gov't securities (assets)
SDRs (liabs)
Total liabilities
total financial assets
total liabilities
Implied market value
B. Alternative calculation of Market Value as in Laitner & Stolyarov less plus less plus Less plus L100, R1 L100, L105, R1 L105, L106, R1 L106, L107, R1 R25 R17 R13 Household s and non profit total financial assets
Househ olds and non profit total liabilitie s
State & local governm ent total financial assets
State & local governm ent total liabilities
Minus L107, R23
Plus L108, R1
plus L108, R15
Equals
Monetary Authority (excl FRB) Total financial assets
Monetary Authority (excl FRB) total liabilities
Total business, L&S definition Implied market value
Fed governm ent
Fed governm ent
Rest of World
Rest of World
total financial assets
total liabilities
total financial assets
Total liabilities
C. Market Value from Identified Components minus plus plus plus equals
Plus
minus
L101,R1
L101,R16 L5,R23
L102, R41
L213,R4
Panels B, C Panels B, C and D and D
nonfinancial business
nonfinancial business
noncorporate business
non-financial non-financial financial financial financial Total corporations business corporations corporations corporations business
total liabilities
value of household equities equity in (including noncorp farm business equities)
total financial assets
market value of US nonfinancial value of equities business
total financial assets
plus
total liabilities
Equals
minus
Plus
L230, R1
L230, R16
total business
total business
Stock of Identified Stock of US foreign direct market value direct of US investment investment in US business abroad
D: Reconciliation of Identified Market Value with Laitner & Stolyarov Market Value minus minus Equals plus plus Plus
equals
L5 row 33
Panel B
Total Identified Assets
L5, R20
Panel C
all sectors
all sectors
Total Liabilities
Assets not included in liabilities (ie equities, gold and Unidentified sdrs) Liabilities
L213, R3
L5, R21
rest of world all sectors identified market value of US Corporate business Equities
gold and SDRs (other assets not included in liabilities)
total business total market value as in L&S
equals
total business Identified Market Value of Domestic Business Capital
F. Capital from BEA nonresidential data (as in Laitner & Stolyarov) Plus TA 4.1, R1 NIPA 5.75A, R1
equals
Private
Private
private
non-residential fixed assets
Inventories
non-residential fixed capital and inventories
Cf Laitner & Stolyarov
"business fixed capital and inventories"
G. Alternative BEA capital measure data excluding non-business non-residential capital but including residential capital less less equals Plus Less Less equals Plus TA 4.1, TA 4.1, TA 4.1, R49 TA 5.1, TA 5.1, TA 5.1, R 14 NIPA r1 R46 R2 R6 5.75A&B L1
equals
private
non-profit institution s
persons
business
Private
non-profit institution s
owneroccupied
business
Private
business
nonresidentia l fixed assets
nonresidential fixed assets
nonresidential fixed assets
nonresidentia l fixed assets
residentia l fixed assets
residential fixed assets
residential fixed assets
total fixed assets
Inventories (end-year)
total reproducible capital
H. Tangible Assets from flow of funds, BEA data and proxies for land Plus Equals Plus b102, r2 b103, r2 Panel M nonfarm non-financial corporate
nonfarm noncorporate
tangible assets (inc land)
tangible assets (inc land)
nonfarm nonfinancial business tangible assets (inc land)
plus Panel N
equals
financial corporations
farms
business
tangible assets (inc land)
tangible assets (inc land)
Tangible assets (inc land)
M: Actual and imputed tangible assets of financial corporations equals
Of which:
Plus
TA 4.1 L 27
Tangible assets
real estate (using ratio to structures from nonfinancials), o/w
Structures, of which
Nonresidential structures, direct from BEA
Plus
Plus
TA 4.1 L 26
Residential structures (set to zero, since Z1 imputes all to nonfinancials)
Land (residual)
nonresidential equipment and software, direct from BEA
Inventories (set to zero because Z1 imputes all to nonfinancials)
N: Actual and imputed tangible assets of farms (corporate and noncorporate) using BEA tangible assets equals of which: plus plus plus imputed TA 4.1 TA 5.1, TA 4.1 L5 imputed L6 L18 Tangible assets
real estate (using ratio to structures from nonfarm noncorporates)
Structure s, of which
Nonresidentia l structures , direct from BEA
residentia l structures (only tenantoccupied)
Land (residual)
nonresidential equipment and software, direct from BEA
residential equipment and software (using ratio for nonfarm noncorporates)
plus NIPA 5.75A L2 Inventories
