N

t

* 

cannot be observed directly but can be inferred from observing the

process by which the housing stock evolves over time.  For example, this

year's stock may be related to last year's stock by the following:

N

N

% (

(

N

(

)

(7)

t

t

&

1

t

&

N

t

&

1

In this traditional  stock adjustment  model this year's stock is equal to

last year's stock (

N

) plus a fraction of the difference between th

t 1

e

desired stock and the actual stock.  Combining terms in (7) and

substituting (6) yields:

P

N

  (

f HH

,

y

,

Z

,

h

%

(1

& (

)

N

t

t

t

t

(8)

P

t

&

1

o

t

Regressing current stock on lagged stock and the presumed determinants of N*

permits estimation of the adjustment rate, 

(

; the effects of exogenous variables on the

desired stock of housing, and the stock of desired housing itself. Having estimated the

parameters of (8), it is possible to estimate the unobservable  desired  housing stock. 

In transition economies, given the constrained supply responses associated with

heavily interventionist policies, econometrically estimated parameters of (8) may be of

interest for forecasting but of less interest for determining  potential demand,  which may

be better estimated by examining relationships between the size of the housing stock for

market economies

 and the determinants indicated in Eq. (6).

It might be assumed, for example, that once market forces are unleashed in a

transitional economy such as Poland's, the housing sector will be increasingly pulled in

ways to replicate the sectoral performance of market economies.  Countries such as

Turkey, Chile, and Malaysia, for example, with income relatively similar to Poland, had

1.20, 0.99, and 1.03 households per dwelling in 1990, compared with a figure of about

1.09 for Poland (Warsaw) in 1990.  International comparisons facilitated by data from

efforts such as the Housing Indicators Project can establish  norms  of such outcomes in

relation to the determinants listed in Eq. (6).   Expected  levels of housing stock in relation

to households and other variables can be established based on estimates of Eq. (6) using

cross country data, assuming that an average

 N

 = 

t

N

t

*.

  Estimates of 

N*

 for Poland so

obtained can be interpreted either directly, as  potential demand,  or can be entered

directly into a stochastic version of Eq. (7) and used as an alternative method to estimate

the adjustment rate, 

(

.

I 10