What kind of construction delays might we see in Q4 2018?

Dec 4, 2018, 14:19 PM by Nathan Adkins

Since 2011, more new units have hit the U.S. apartment market than in the previous year. This seven-year acceleration looks to be coming to an end, however. In competitive markets, what goes up must come down. Recent construction data suggest that multifamily supply growth will peak in Q4 2018 and correct to near the 10-year average by 2020.

Figure 1. Apartment Completions Forecast: EA Sum of Markets

181204 construction delays - fig 1

Source: CBRE Econometric Advisors, Q3 2018.

If you’ve been reading EA’s Q3 2018 Apartment Overview and Outlook, you may be saying, “don’t completions routinely come in at 15-25% under the construction projections (Fig. 1’s blue columns)?” What kind of construction delays might we see in Q4 2018?

Our two-year forecasts use property-level construction data from Dodge Data & Analytics. Dodge tracks U.S. construction projects through six phases of construction, from pre-planning through completion. The tracking data include the number of units and the target completion date for each project. These data allow us to forecast near-term supply growth with great precision. Put simply, the completions forecast for the next two years come directly from observed building projects; they are not predictions from a statistical model. Beyond those two years, construction data are less complete and less reliable, so our statistical model kicks in to estimate supply based on modeled supply and demand dynamics.

Our near-term supply forecast is precise (in that it is an observation, not a prediction), but that does not make it necessarily accurate. Dodge reports target completion dates for thousands of properties across the country. For a variety of reasons, projects are not always completed on schedule. Things like labor shortages, inclement weather, and shipping delays for foreign-made construction materials all contribute to late deliveries.

Figure 2 shows the percentage of forecasted completions, since 2014, which arrived on time by the number of quarters forecasted ahead. On average, 78% of completions forecasted for next quarter arrive on time, while 22% are delayed to subsequent quarters. This pattern of over-estimation continues for 3 quarters, pushing more and more near-term completions to later quarters, which haven’t yet begun to forecast delays, and so are then under-estimated.

Figure 2. Completions Forecast Overestimation

181204 construction delays - fig 2

Source: CBRE Econometric Advisors, Q3 2018.

Using the percentages from Figure 2, we can adjust our construction forecast to reflect historic over- and under-estimation. Figure 3 compares the adjusted and original construction forecasts. The adjusted completions are now forecast to peak in Q2 2018 and experience a slightly slower deceleration in the near term.

Figure 3. Sum of Markets Completions

181204 construction delays - fig 3

Source: CBRE Econometric Advisors, Q3 2018.

We can identify a pattern of delays in recent completions data at the Sum of Market level, but at the market and submarket levels, data on delays are highly variable. Adjusting the data comes with the risk that we will bias our forecast by smoothing over idiosyncratic, but accurate, near-term market behavior. Finally, a parsimonious model is often preferred in forecasting because it is less prone to overfitting errors and is more interpretable. In this case, we use construction data because it is precise and has a direct connection to the market reality, though delays may occur. These are actual building projects currently under construction with expected completions dates in these quarters. Finally, we don’t mind some variability in the exact timing of near-term completions because the construction data accurately predict cumulative completions over the next two years. Figure 4 shows the average percent of projected completions which complete within up to 8 forecasted quarters.

Figure 4. Average Forecasted Completions Delivering within Timeframes

181204 construction delays - fig 4

Source: CBRE Econometric Advisors, Q3 2018.

By using two years of construction data and then shifting to the statistical model, we incorporate precise near-term data into rent forecasts while also maintaining 2-year accuracy in total supply. That is, while next quarter’s completions projection is likely too high, yearly and bi-yearly projections are increasingly accurate.


Deconstructing CRE

172 posts

Last post : 07/08/2020

redirect pin user minus plus fax mobile-phone office-phone data envelope globe outlook retail close line-arrow-down solid-triangle-down facebook globe2 google hamburger line-arrow-left solid-triangle-left linkedin play-btn line-arrow-right solid-triangle-right search twitter line-arrow-up solid-triangle-up calendar globe-americas globe-apac globe-emea external-link music picture paper pictures play gallery download rss-feed vcard