Re: Controlling Inflation in Capital Asset Prices

["Per Gunnar Berglund" <pgb@xxxxxxxxxxxxxxxxxx>]

In reply to John Gelles:

WARNING! LONG POST!!

-----

John Gelles wrote (in part):
> I know I am late in this discussion and that Per, Warren, and Bruce, etc.,
have made lots of points. But I would not object to a summary of where the
discussion stands and where it is heading.  <

Per:
Let me try to sum up where I stand on this:

1. DEFINITION OF CAPITAL COST
I figure the easiest way to get at this is to think in terms of an ordinary
loan. The "cost" (speaking loosely here) of the loan consists of two parts:
interest and amortisation. Now, the interest part is usually _defined_ as
the amount the debtor must pay in order to keep the principal intact.
Suppose the debt is D(0) at the beginning of the year, and D(1) at the end
of the year, and that the debtor has transferred the amount T during the
year. Then the interest is R = T + D(1) - D(0) .

Now, the problem here is that the principal is usually thought of in terms
of money value. But suppose that assets in general have increased in price
by e.g. 10% during the year. If D(1) = D(0) when measured in terms of money,
the creditor will have LOST 10% of the real value of the debt during the
year. That's the hangup. We need a definition of interest which is based on
maintaining intact the REAL value of the principal, not the nominal value.

If we assess the real value of the principal by using the Asset Price Index
I have been suggesting as a deflator, the interest equation will be slightly
modified to REAL interest RR = T + D(1)/P(1) - D(0)/P(0), where P(0) and
P(1) denote the values of the API at the beginning and end of the year,
respectively.

This is how I perceive capital cost should be defined. According to the
principles of alternative cost, all assets will have a capital cost
corresponding to the real interest, as defined above, on the principal (or
exchange, or "present") value of the asset. Capital cost thus defined will
enter into the total cost of production.

2. EXCHANGE NATURE OF CAPITAL COST
Capital cost is relevant not only to the assets subject to lending and
hiring, but also to the "owner-occupied" parts of the aggregate wealth. For
the latter category, an imputation of capital cost is generally needed,
since there are no actual transactions carried out between the owner and the
user of those assets (both being the same person).

For economic _decision-making_, however, the actual distinction between
owner and user, debtor and creditor, and the actual transactions occurring
between the two, are of paramount importance. We may regard a non-zero
capital cost as a kind of "tax" (the revenue of which accrues to the
creditor) on credit (hiring, lending) arrangements. Obviously taxes will
tend to diminish the volume of such transactions, a restriction which will
tend to reduce the efficiency of the resource allocation: the right capital
object might not reach the right user, so to speak. The higher the rate of
"taxation", the graver this problem will be. Interestingly, the problem of
misallocation will also obtain whenever the "tax" goes BELOW the zero level,
meaning that credit arrangements are "subsidised". By standard microeconomic
reasoning, we may conclude that the zero level of "taxation" is the optimal
one as far as the allocative aspects are concerned.

Therefore, I have argued, the monetary policies should aim at keeping the
aggregate level of capital costs at the ZERO level at all times. In terms of
the real rate equation above, this means setting RR = 0, implying that T =
D(0)/P(0) - D(1)/P(1).

3. CAPITAL ASSET PRICES AND CAPITAL FORMATION
The basic reason why we would want to control the asset price-level is (1)
to control capital costs as a component in costs of production, hence
stabilising the general level of output and employment; (2) to dampen the
swings in capital formation. These two aspects are certainly intertwined.

Keynes's Marginal Efficiency of Capital (MEC) approach to capital formation
was based on an idea of Irving Fisher's, namely that the prospective
quasi-rents from a capital good should be discounted to a present-value,
which may or may not exceed the cost of reproduction of that kind of good.
The modern version of this line of reasoning is known as "Tobin's q", which
compares second-hand market prices of capital goods with the corresponding
cost of new production. The higher the ratio of the former to the latter
(i.e. the higher Tobin's q), the faster the rate of capital formation is
supposed to be.

I believe there is much truth in this approach, but there are also some
elements of non-truth. The discounting procedure Keynes and Fisher had in
mind stretches over the whole life-span of the capital good in question.
This is a very doubtful procedure, I think. Most capital formation takes
place in a speculative setting, where the expectations of the exchange-value
of the capital good in spe _at the time it is finished and can be sold_ are
a factor of major importance. The Keynes-Fisher picture of long-term
expectations and discounting procedure may not be very relevant when capital
goods will be subject to re-sale within a few months or a couple of years at
most.

This is the chief reason why I find it so important to control not only the
LEVEL of prices of capital goods, but also the RATE OF INCREASE of this
price-level. A high price-level on second-hand capital goods will stimulate
investment, but only if this price-level is expected to last for at least a
couple of years. If there is a fear of lowered asset prices the next few
years, a high present price-level may not help up capital formation very
much. Conversely, a rather low price-level of second-hand capital goods may
well combine with a high level of capital formation, if there are firm
expectations that the prices will go up sharply in the near future.

By adopting a policy strategy of keeping r = g(API), there will be no
GENERAL opportunity to make excessive capital gains -- nominal interest
costs will swamp the nominal capital gains. (of course some may be lucky and
have capital gains in excess of r, but this will only mean that others have
bad luck and suffer losses due to capital gains being less than r. These
categories will always cancel out in the aggregate if the r = g(API) policy
is kept up.) Therefore, the speculative element in capital formation will
vanish -- only the LEVEL of capital asset prices will be relevant. In other
words, this policy will put us back to the world Keynes and Fisher, in an
artificial manner. The MEC and Tobin's q will again start to function as
they were supposed to.

4. PASSIVE ADAPTION OF r TO g(API)
The policy of keeping r = g(API) must not be mistaken for an attempt to
directly control the API (even if this may be necessary at extremely
depressed states of the economy, corresponding to the schoolbook Keynesian
"liquidity trap"). The normal procedure will be to let the API go where it
will, and to adapt r ex post to the g(API). Of course the central bank may
opt for setting r slightly higher than the g(API) if there are good reasons
to expect an acceleration of the latter in the near future. But basically,
capital asset prices should be allowed to go whereever they will, and the
rate of interest should take a more passive "back seat" role in the
interplay.

It should be noted that even such a passive adaption would tend to feed back
on the API. There will be no great instability in the g(API) since the
market knows that any increase in g(API) will lead to raised interest rates.
So, I reckon there will be a great deal of "self-containment" in the market
once this kind of policy is launched.


All best,
Per


Per Gunnar Berglund
Lilla Sallskapets vag 60
127 61  SKARHOLMEN
SWEDEN

Voice/fax +46-(0)8-883065