1 These statistics on building construction authorized by local building permits measure building activity in all localities having building-permit ystems-rural nonfarm as well as urban. Such localities (over 7,000) include about 80 percent of the nonfarm population of the country, according to the 1950 Census. The data cover both federally and nonfederally owned projects. Figures on the amount of construction contracts awarded for Federal projects and for public housing (Federal, State, and local) in permit> issuing places are added to the valuation data (estimated cost entered by builders on building-permit applications) for privately owned projects;

construction undertaken by State and local governments is reported by local officials. No adjustment has been made in the building-permit data to reflect the fact that permit valuations generally understate the actual cost of construction, nor for lapsed permits or the lag between permit issuance or contract-award dates and start of construction. Therefore, they should not be considered as representing the volume of building construction started. Components may not always equal totals because of rounding. 2 Revised.

The data shown here do not include temporary units, conversions, dormitory accommodations, trailers, or military barracks. They do include prefabricated housing, if permanent.

These estimates are based on (1) monthly building-permit reports (adjusted for lapsed permits and for lag between permit issuance and the start of construction), (2) continuous field surveys in nonpermit-issuing places, and (3) reports of public construction contract awards.

Beginning with January 1954 data, the estimating techniques for the privately owned segment of the housing starts series were revised to combine (1) a monthly reporting system expanded to include almost all buildingpermit-issuing localities (accounting for nearly 80 percent of total nonfarm population), with (2) a newly designed sample of counties that permits more efficient operations and a greater degree of accuracy than previously. The new series is continuous with statistics for earlier dates except that the urban and rural-nonfarm distribution shown previously is replaced by metropolitan-nonmetropolitan and regional estimates. Data on type of structure (1-family versus rental-type structures) are continued from the old to the new series, and are available on request.

The error in the total private nonfarm estimate due to sampling in the

nonpermit segment is such that for an estimate of 100,000 starts the chances are 19 out of 20 that a complete enumeration of all nonpermit areas would result in a total private nonfarm figure between 98,000 and 102,000. For metropolitan-nonmetropolitan or regional components, the relative error is somewhat larger.

Data by urban and rural-nonfarm classification for periods before January 1954 are available upon request. Annual metropolitan-nonmetropolitan location data not available before 1950; monthly figures not available before 1953; regional data not available before January 1954.

Private construction costs are based on permit valuation, adjusted for understatement of costs shown on permit applications. Public construction costs are based on contract values or estimated construction costs for individual projects.

Housing peak year.
Less than 50 units.

• Preliminary.

Revised.

'Not yet available.

U. S. GOVERNMENT PRINTING OFFICE: 1956