Do Permits Lead Starts?

“Building permits, an indicator of future building activity….”
--Wall Street Journal, March 17, 2004, p. B2

It’s a common notion that building permits are a predictor of future housing construction. After all, homebuilders need to obtain a permit before starting construction, so the number of permits issued should serve as a reliable indicator of future activity, right? However, there are several reasons why permits are a weak leading indicator of new construction.

Lead is Short

The reliability of permit issuance as a predictor of housing starts is confounded by the extremely short time between issuance and start. The average amount of time between permit issuance and start for one-family homes is about three weeks. Even though tract developers may acquire all their permits when development begins and then stagger new construction over time, most one-family housing starts occur in the same month that the permit is recorded, Exhibit 1 shows.

Furthermore, U.S. Census Bureau studies find that a small number of starts—about 3 percent—occur before a permit is issued; some jurisdictions allow ground breaking to occur before the final permit is issued. Also, housing starts data are based on a calendar month, while some permit-issuing jurisdictions close their permit-recording books a few days before month-end, causing the permit to be recorded in the following month. Finally, some starts escape recording.

Sample Differences

The ability to forecast permits from starts is also blurred by differences in characteristics of the permit and start samples. The major difference is that starts can occur outside of permit-issuing jurisdictions. To illustrate, one-family housing starts have exceeded permits by 5 percent over the past 9 years (starting in January 1995, the Census Bureau expanded its sampling frame to include all 19,000 permit-issuing jurisdictions in the United States). Thus, permit data may occasionally be a misleading barometer of new construction if the amount of activity outside of permit-issuing places suddenly changes.

To test statistically whether permits lead starts, the monthly change in seasonally adjusted one-family permits was correlated with the change in seasonally adjusted one-family starts for the same month, and for each subsequent month. Over the past 20 years, the simple correlation coefficient for the concurrent month was 0.62, but it was insignificantly different from zero when starts led. That means there was a close relationship between permits issued and housing starts in the same month, but not between permits and starts in subsequent months.

Exhibit: 1 Most Starts Occur the Same Month as Permit Issuance

What About Multifamily?

For properties with two or more dwellings in the structure, the average lag between permit issuance and start generally is twice as long as for one-family dwellings. Further, virtually all multifamily housing is built in permit-issuing areas, and compliance with housing authorization laws is more universal. This should lead to a better linkage between permits and starts; indeed, since January 1995, multifamily permits have exceeded starts by a scant 1.5 percent, the opposite of the relationship for one-unit construction. The fact that only about one-third of multifamily starts occur within the same month as the permit also suggests a potential for causality from permits to starts.
Correlation coefficients between multifamily starts and multifamily permits, for the coincident and lagged months, show a weaker but still very significant correlation of 0.32 for the coincident month, but correlation coefficients that were not statistically different from zero for lagged permits. In other words, the results were comparable to those for one-family construction: There was a close relationship between permits and starts in the same month, but not between new starts and permits issued in previous months.

Are Permit Data Valuable?

Absolutely! Although permits may not explain future construction activity, permit data are valuable for a number of purposes. Current-month permit data serve as a confirmation that the starts’ reading is not aberrant. That is, months in which starts and permits move in opposite directions indicate a mixed signal, whereas the economic signal is reinforced if both series move in the same direction. Further, permit data are available at small geographic levels, such as county or metropolitan area, which make them useful for understanding local building conditions.