[OPE] review of Dumenil & Levy's Capital Resurgent

From: Jurriaan Bendien (adsl675281@tiscali.nl)
Date: Sat May 31 2008 - 06:39:09 EDT

That's a useful review. My own take on this which I aim to write up some time is that Dumenil & Levy's empirical measurements of productivity, based on a relationship between official data on inputs and outputs, can be questioned conceptually. Once this is done, their notion of productivity is undermined.

Among other things, Marxist and non-Marxist economists often assume that data on net output (or, roughly, the sum of factor incomes) measure the value of new products produced, whereas net output only intends to measure the new gross value added by production. 

It may be that if we deduct total intermediate consumption from total gross output of production (the total turnover or total gross sales revenue) defined in some suitable way, that we obtain an approximate measure of the value of all the new products created together within an accounting interval. 

But, if let's say a factory produces a car which sells at the factory gate for $20,000, that $20,000 unit price does not simply represent the value added or the factor incomes generated by the factory - it also includes intermediate operating costs in building it. When you buy the car, in other words, you do not just pay for the wage costs and gross pretax profit income of the manufacturer implied in building the car, but also the intermediate operating costs, including for all the componentry purchased to make the car. In Marxist phraseology, the unit price of the car is not V+S, but C+V+S.

In the NIPAs, unlike UNSNA, intermediate consumption is not tabulated in the derivation of gross output (as is the practice in UNSNA), but it is shown in the input-output tables. Taking the (rounded) data for the year 2004 as example,

Gross output = $21,346 billion
less intermediate consumption (= $9,612 billion)
equals net output (or GDP) = $11,734 billion

So in the NIPA's, total intermediate consumption (value of purchased goods and services inputs used up in production) is about 45% of total gross output (roughly, the total turnover). About half of total intermediate consumption in the US consists of services. The largest intermediate items by value are manufactured goods and financial/business/professional services.

Why is all this significant? For a number of reasons, of which I will mention a few obvious ones. 

As said, the new value added by a sector does not equal to the sum of unit prices of products actually sold by that sector; inputs and outputs are valued at producer's prices. 

Secondly, the magnitude of intermediate consumption can fluctuate semi-independently from the magnitude of net output. This can be easily verified in the NIPAs since the statisticians in fact calculate how much total inputs it takes to produce one dollar of output by sector. 

Thirdly, if the nature of throughput and output changes a lot, qualitatively, across a couple decades, then it becomes difficult to compare earlier and later financial data because it may refer to completely different objects. If for example in 1980 the output of phones was (say) $10 billion, and in 2008 the output of phones is $20 million, how do we evaluate the increase in productivity since mobile phones were only a niche market in 1980 and produced very differently? 

Fourthly, economic accounts provide no measures of physical productivity - the market value of products produced may go up or down even although the physical output keeps increasing strongly. 

Fifthly, as regards modern companies, the total revenue of the enterprise may bear no relationship to the total unit prices of outputs sold, because the enterprise has significant costs and revenues completely unrelated to those outputs.

The accounts are built up on the principles that for every selected purchase there is a sale, for every selected expenditure there is an income, for every selected asset there is a liability and so on. But as regards capitalist production, more value comes out of it than goes into it (otherwise there is little point in producing), and nearly half the outputs of some enterprises are the inputs of other enterprises. The account then tries to straddle the difference between capital accumulated and total new product values with a bunch of distinctions which aim to define and measure the net addition to wealth. This results in the notion of value added, but this notion in reality does justice neither to accumulated capital nor to the actual value of products produced and sold, except that we can then estimate that for total production, valued in some suitable way, there is an aggregate standard price (=GDP) equal to the value of the total new products produced and circulated in an accounting interval. 

In summary, I think the "fetish" people often fall victim to is that the sectoral "net output" values cited in economic statistics represent the sums of the "unit prices of products and services produced". On that basis, we can of course devise all sorts of ways to measure productivity by relating sectoral inputs and outputs, but, if in reality those inputs and outputs merely describe a predefined subset of total costs and total sales, we ought to be more skeptical of the productivity indicators provided.

I think that properly considered, Dumenil & Levy's conclusion ought to be that although physical productivity increased strongly, the actual income generated by that additional physical productivity was in no proportion to the increase in physical output. That would be a more Marxian conclusion: more and more commodities are produced, and more are produced per worker, but their unit-values declined in real terms and therefore the real incomes generated by them was not proportional to the increase in physical output. But this is obviously difficult to test, if we lack data on outputs of physical product units. This conclusion means that productivity growth did not stagnate in labour terms or physical terms, but that the longterm tendency was for output to be devalued in real terms.


ope mailing list

This archive was generated by hypermail 2.1.5 : Mon Jun 02 2008 - 00:00:06 EDT