State data on

STATE DATA FREQUENTLY SOUGHT... BUT IS A SOURCE ITEM
housing starts is in constant demand--from all types of users, and for a wide
variety of purposes. Despite the demand, technical difficulties have prevented
the satisfactory development of such data by either official government, or
private sources. For a few years, during the mid-1950's, the Bureau of Labor
Statistics provided estimates for a list of selected states. For a number of
reasons, the series was discontinued.

In recent years, those needing state estimates have been forced to rely
primarily on building permit data. However, comparisons of state permit data
with year-built data from the 1960 Census vary widely and reaffirm doubts
that permit totals reflect the probable level of construction volume.
This is mainly due to an absence of permit systems in many areas (mostly non-
metropolitan) and possibly also to lax enforcement in some areas where permits
are technically required.

In California, Maryland, and Hawaii, for example, the permit levels
appear to fairly represent the levels of new construction. On the other hand,
some states had apparent construction levels as much as three, four, and five
times the reported level of permits.

...

MAIN USES For precise market analysis requirements, these estimates
should be regarded as unsatisfactory. However, for such purposes as estimating
the impact of home building on a state's economy, the estimates should provide
a satisfactory working basis. Likewise, for builders, manufacturers, distrib-
utors, and other contemplating multi-state activity, they should provide a
more reasonable basis for making comparisons of the relative level of building
activity among various states--more reasonable, that is, than data heretofore
available. Likewise, for firms already engaged in multi-state activity, they
provide a reasonable basis for grouping states into special purpose regions.
By the same token, the data can be used to advantage by national firms in
dividing up sales territories, establishing sales quotas, establishing distri-
bution points, establishing and evaluating inventory controls, etc. While
they fall far short of desirable reliability, they would nonetheless seem
superior to the "best guesses" on which many important decisions have necessarily
been based in the past.

A NOTE ON METHOD ... "Year-built data" from the 1960 Census was
adjusted to official census estimates of total housing starts for the 1959-
1960 period--assuming a consistent state-by-state percentage of overstatement
on the number of housing units recently built. (for the benchmark period)

This permitted estimates of starts levels for each state which could be used with state permit data to derive an adjustment factor. Adjustment factors are then applied to state permit levels for each of the years estimated.

FUTURE REFINEMENTS ... A number of refinements in this data are both possible and desirable--and these will be developed as time permits. In particular, it is desirable to differentiate between sale housing and rental housing--or between one-family and multi-family housing. This distinction would be important for many users. Also, because the adjustment factors were derived from totals, there are undoubtedly some biases in the starts estimates for states which have a low rate of permit coverage in combination with recent major shifts in the mix of multiple housing. This is because the major discrepancies between permits and actual construction occur largely in one-family home structures--the dominant type in non-metropolitan areas without building permit systems.