Re: nat'l accounting...

["Bruce R. McFarling" <ecbm@xxxxxxxxxxxxxxxxxxx>]

On Fri, 29 Aug 1997 15:44:52 +0200, "Per Gunnar Berglund" 
<pgb@xxxxxxxxxxxxxxxxxx> wrote:

>    The transformation finally occured! Dr. McFarling turning into Mr.
>McCarthy, accusing me of "non-Keynesian activities". Well, I will leave 
>it up to others on this list to vote "guilty" or "not guilty".

	I'm merely trying to find the post Keynesian economics in it. 
I'm trying to figure out which Post Keynesian economists are left. 
Anyone coming from the American Institutionalist tradition(s), such 
as JK Galbraith, is obviously to be dropped out, since they would 
never accept that the 'transaction principle' used to define National 
Income accounts be used to delimit the scope of economics. Not to mention 
the neoclassical conflation of microeconomics and utility maximising models. 
It seems as if Paul Davidson is out, if we are to put the construction of 
time-invarient measures as a criterion for any macroeconomic aggregate, 
and also if we are to ignore the sequence of obligations and returns 
at the firm level between the hiring decision, management of production, 
and earning returns in the market.  Kaleckians seem to be out of it 
because of their Marxian foundations.  Joan Robinson stands (or lies, 
if we wish to be precise) accused of being intellectually lazy for 
her argument on the epistimological status of economics.  Sraffa has 
nothing to say about the evaluation of the capital fund.
	I've been trying to figure out which Post Keynesian economics 
this all connects up with.  I thought there was a point of contact 
with the insistence on observables in the national income accounts, 
but that turned out to be common language referring to quite different 
things.


>    But let me, as a kind of dessert to this main course, offer this
>delicious recipe, which might give some food for thought, if not the
>physical nourishment that seems to be the chief occupation Dr. McFarling and
>others:
>    (1) "You cannot add apples and pears". Hence intertemporal comparisons
>are meaningless;

	I will repeat. You cannot directly compare apples and pears.  Any 
conversion of apples and pears to permit comparison will either sacrifice 
precision or accuracy.  We can compare apples and pears as small fruit (loss
of precision, maintenance of accuracy) or we can convert apples to "pear 
equivalents" and compare them (loss of accuracy, maintenance of precision). 
When we aggregate direct observables, we will ordinarily have an increase 
in heterogeneity in the amounts that we aggregate.
	It appears to me that youir response to this increase in heterogeneity 
is to claim that any terms that are not homogenous cannot be compared *at 
all*, and therefore we must necessarily make believe that we have homogeneous 
classes.  You may describe this in terms of imposing the homogenity with the 
measure that you use, but a measure cannot impose homogenity, so that is, in 
my mind, the same thing.

>    (2) "You cannot compare Mr. Smith and Mrs. Jones". Hence interpersonal
>comparisons of well-being are meaningless.

	Interpersonal comparisons of well-being are meaningless under 
an individual utility maximising framework.  You are the one who has 
specified the reliance on an individual utility maximising framework. 
That, in and of itself, does not guarantee that you are *not* a Post 
Keynesian, since the label appears to fall on both side of the line. 
But you can't have it both ways: you have to use it validly if you 
use it.

>    Voilà! Le coup(e) von Mises--McFarling! Un bomb sûrprise!
>    Warning! This dish may cause serious indigestion, of the kind that is
>epidemic amongst libertarians.

	Am I supposed to be concerned that I would cause indigestion if 
mixed with von Mises?

>Per Gunnar Berglund reply:
>    Lovely. Let me then ask _exactly where_ the limit goes between
>"meaningful" and "meaningless" comparison? Ten years, one year, one month,
>one week, one day or maybe one hour?

	This is exactly the point I made regarding fuzzy set theory. 
There *is no exact boundary* between meaningful and meaningless 
comparisons when comparing heterogenous classes.  If you pin down 
one side of the comparison (say, the present), the *degree to 
which* the comparison is meaningful declines as we span a longer 
period of history.  Just as the degree to which the comparison is 
meaningful between different contemporaneous economic systems 
will vary.
	"The meaningfulness of comparisons has a crisp-set membership 
function" is *your* assumption.  What argument do you present in 
its favor?

>    Over a one-year period, the change in output composition is highly
>likely to change substantially. Why should this change be tolerable, while
>the 10-year change is not?

	I would argue that, while the individual quantities per unit time 
in the output vector may fluctuate quite a bit in a year, the composition 
of categories is far more stable.

>    You have clearly missed the point with the _chain_ index property of the
>volume concept. You reason as if we were to use the same weights over the
>whole period, thus having a _fixed-base_ index, instead of a chain index.

	No, I do not.  You reason as if we can render the things we measure 
conmesurate simply by the construction of the measure.

>...And in order to get a clear-cut and unambiguous index, we should have 
>dt time periods (continuous time) and also dt continuous re-weighting.
>That is the theoretical volume concept.

	But what is your argument regarding why we should aim for a 
clear cut and unambiguous index?  Does a clear cut and unambiguous 
index eliminate the problems involved in comparing heterogenous 
aggregates?  Of course it doesn't.  Provided it is as accurate 
as another index, it would seem to be preferable.  But do I 
understand correctly that you are claiming that the selection of 
an index should be on these grounds, and not on the grounds of 
empirical accuracy?


>    Again, and this is for the third time: The concept of economic volume 
>is NOT a concept of physical volume. If you fail to see this crucial
>difference, I am afraid you are beyond all rescue.

	Understood.  Your concept of economic volume is not a concept of 
physical volume.  As I said, I agree that you can't from your aggregate 
volume measure to physical volumes.  I am just asking for an argument 
which explains why it is a good thing that we abandon phsyical volumes 
for your measure.

>    Your argument here is basically part (1) of the Coupe I suggested above.
>Indeed you explicitly say that "the actual GDP can only be determined in
>nominal terms". Hence, intertemporal comparisons in terms of _economic_
>volumes are meaningless. Of course, there will still be scope for
>intertemporal comparisons in nominal terms, but they are really not very
>useful.

	I apologize if I said that actual GDP can only be determined in 
nominal terms.  I should have said that actual GDP is a monetary value. 
Is that better?  Of course, the physical output can be compared to the 
extent that it can be compared.  How does your 'economic volume' permit 
us to compare the physical output to a greater extent than that?

>    Here you clearly admit that even if all other magnitudes are nominal, we
>must still be able to define the _volume_ of employment. You are clearly
>self-contradictory here, since you have stated above that any summation of
>non-homogenous magnitudes is meaningless.

	You insist on reading this in.  Although it is your position regarding 
what I have argued, it is simply not what I argued.  My arguemtn is that 

	(1) A meaningful *empirical* basis of comparison must be found, 
if the measures that we construct are to have any meaning.  I will assume 
that you agree on that, on the basis of your earlier arguments regarding 
the exchange principle (though I admit that I still have trouble 
believing that someone would actually want to abandon all economic 
reasoning that does not fit into a SNA framework)

	(2) Keynes concept of 'ordinary labor' (combined with Commons concept 
of strategic and complementary resources) can provide us with a meaningful 
basis of comparison.  That is, 'ordinary' labor involves skills that are 
sufficiently commonplace or can be acquired sufficiently quickly that
managers 
don't worry about whether a specific applicant possesses the specific set of 
skills required for the task.  This is the situation in which labor is the 
most like a complementary resource, and least like a strategic resource, 
and involves the type of labor that can be performed by the broadest cross- 
segment of the labor force.  Since workers any labor category possessing
skills 
or other strategic resources include large numbers who can also fill the role 
of 'ordinary labor', their earnings in comparison to the earnings of ordinary 
labor provides a meaningful comparison of the degree to which their labor is 
a strategic resource.  it is a matter of common knowledge that if managers 
could replace them with ordinary labor at an equivalent wage cost and acheive 
the same effectiveness, they *would* replace them with ordinary labor, and 
therefore the wage multiple gives a conservative measure of the bargaining 
position of special labor compared to ordinary labor, for that time and
place.  
Clearly when we convert using these coefficients and aggregate, it may be
argued 
that the aggregate is biased, and it could be argued that the labor unit will 
probably understate the value we would arrive at if we could make a direct 
comparison.  But the labor unit has the advantage that it is based on
observables 
(where the observables include manager's preferences for hiring labor with as 
little strategic power as possible).


>Nonsense! (sorry): The "intertemporal invariance" you have in mind, is
>clearly the Hicks-aggregation notion of an unchanged index-number "basket".
>This completely misses the chain-index nature of the concept of economic
>volume. (Vide above)

	I will argue that a chain index does not solve the problem of 
comparing heterogenous groupings in systems that are evolving.  I have 
argued that a chain index does not solve this problem.  You assert that 
it does, and your argument appears to be that I would agree with you if 
I understood that you are talking about chain indices.  Well, I understand 
that you are talking about chain indices, and I still do not agree with you.
Do you have an argument to support your position, or not?

	That is, is it a paradigmatic position that you are taking, that 
construction of a chain index serves to render conmensurate heterogenous 
groups in evolving systems, no matter what evolution the system in 
question is experiencing?  Or can you present some other foundation, 
on which you can build an argument that leads to this conclusion?  Or, 
is there some crisp-set criterion for the types of system evolution which 
can be bridged by a suitable chain-index number, and the types of system 
evolution which cannot be bridged in this way?

>...
>    Neoclassical or not, this brings us back right back where we started
>from. You know, in one of my very first posts on this thread, I stated that
>there are tremendous problems in the actual measurement of the volume of
>input labour services. These problems are not really on the conceptual,
>theoretical level, but on the practical measurement plan.

	Another way to put this is that you claim here that the problems 
are not on the conceptual, theoretical level.  So, OK, none of this is 
a problem.  But don't let that prevent you from stating the conceptual, 
theoretical arguments.

>Therefore I suggested that we should NOT aim at actually performing this 
>measurement, but that we should erect an axiomatic system of lingual links 
>between the output classes and the input classes, links which would enable 
>us to measure the input categories indirectly by our measures of the output
>categories.

	My guess is that this type of approach is exactly why Keynes 
thought it was necessary to have a direct measure.  Since, even if you 
construct your linguistic links, and you have a valuation based on the 
output measures which you argue is effectively tracking the input categories, 
*how do you verify it* unless you have a direct measure for comparison?

>    My simple axiomatic statement was, in the labour services case, that L =
>bY, the "Leontief" specification thus. By measuring Y in volume terms, using
>the chain-index notion of economic volume, we would indirectly get an
>estimate of L in volume terms, by simply adjusting for the differences in
>nominal wage bill and nominal output value in the base-year (the 'b'
>adjustment coefficient).

	If you are to avoid the fallacy of misplaced concreteness, how do you 
verify that the indirect value that you label L measures what you claim it 
measures?  And, even if it provides some success in measuring what it claims 
to measure, comparisons over time will obviously suffer a reduced degree of 
meaningfulness as the span of the comparison is increased: how do you detect 
the loss in accuracy in the comparison?  No measure can measure its own 
measurement error -- how can you determine the measurement error of your 
'indirect measure' unless you can provide an empirically identified measure 
to use as a benchmark?

>...
>    Let me finally add that the notion of volume measurement does indeed
>follow from "neoclassical" or "microeconomic" (or whatever) notions, namely
>the notions of constrained utility maximisation (and/or effort
>minimisation).
>    The mere mathematical method employed, the Lagrange multiplier method of
>constrained optimisation that is, directly yields some powerful insights
>about the intra-personal part of the matter:
>    (a) the individual will spend his income on different goods in
>proportion to their relative contribution to the utility of that 
>individual;
>    (b) the individual will spend his efforts on different occupations in
>proportion to their relative contribution to the work income of that
>individual.

	Constrained utility maximising is not a 'mere mathematical 
method'.  It is a scientific, verifiable model of certain physical 
phenomena that was developed in the 1800's, and borrowed in the late 
1800's by the founders of neoclassical economics.  The analogy that 
was drawn between human behavior and the physical phenomena that the 
model actually applies to turned out to be a false analogy.  Of course, 
there are a variety of ways to react to the problem of having a false 
analogy as the basis of your theory.  One of these seems to be focusing 
on the parts of critiques that are answerable, and hoping that the 
parts that are not unswerable will be lost in the confusion.[*]
	And as a result, some parts of the neoclassical model have 
received a great deal of attention, indeed.  One is the question of 
interpersonal utility comparisons.  You can't do it.  So don't 
be surprised if I ask you to present your argument.
	(1) Why should we accept a measure of welfare based on 
a utility maximising model? and
	(2)  How do you propose to use the two marginal relationships 
you indicate above to compare the welfare of person A and person B?

	Feel free to take more time with your reply to this message.  
Less haste may mean more speed, and I realise that you are participating 
in a variety of subthreads in this thread, so that fitting you answers 
within the three-per-day and average of (??? Ric, how much is it ???) 
per week guidelines must be difficult.

Virtually,

Bruce McFarling, Ourimbah, NSW
ecbm@xxxxxxxxxxxxxxxxxxx

[*] OK, so that's a bit over the top.  Truth be to tell, if people 
are published for answering the answerable bits, there may well be 
enough of that going on that those defending the flawed analogy 
actually believe that the most serious problems have been, or are 
in the process of, being solved.  And, as I have argued at length 
in this forum in the past, the model is not in applicable to *every* 
microeconomic problem, simply inapplicable to *enough* microeconomic 
problems to form an inapproriate basis for macroeconomic reasoning.