Since the beginning of the Pandemic we’ve known there will be declines in construction spending due to #1: Projects halted and temporarily on hold, and #2: a huge reduction in the amount of new construction starts. That’s the easy part. The difficult part is calculating how much within the spending decline due to each cause is being counted in the most recent actual spending and the current forecast.
Although we may never know precisely, this is an attempt to identify how much of the declines in spending are due to #1 and #2. We can set up a model to calculate declines from lost new construction starts (#2), but no source is tracking the amount of projects put on hold (#1), for how long and how many of those projects become permanently shut down.
By comparing the amount of spending declines caused by lost starts to actual spending declines, we get a difference that could be identified in part as being caused by spending declines due to projects temporarily shut down. But that will have a distinct pattern to it. Initially we should see actual spending declines were greater than would have been caused by just the loss of starts, so the difference would be negative. Eventually, projects that were on hold start back up and we should see actual spending declines are not as great as the losses from new starts would indicate. This is caused by a revenue from a few months of previously on-hold projects extending the delayed spending to a later (current) point in time.
Finally, after all on-hold projects have restarted and been reabsorbed into monthly spending, the impact of delayed projects diminishes and the percent spending declines calculated by the loss of new construction starts should more closely reflect the actual declines in spending.
The following cash flow model may not be precise, but it is accurate in its representation of impact caused by lost new starts, and therefore allows to make the comparison noted above. This is strictly for nonresidential buildings.
HINT: Large view, Right Click on table – Open in a New Tab. Thank me later.
This cash flow schedule tracks reduced starts from 2019 through 2020. All other previous months are considered at 100% of the pre-pandemic high in Feb 2020. This sample uses $10bil/mo of new starts as 100%, the high in the 1st quarter, carried out over a 20mo schedule. If the rate of starts were to remain constant at $10bil/mo, then the spending would also remain constant at $10bil/mo. The amounts carried for all months 2019 and 2020 represent the percentage of actual starts recorded, measured as a percent of previous high, the 1st quarter 2020, so $6.6 bil in May is 66% of the pre-pandemic highpoint, February, which here is $10bil.
Notice although starts are forecast to increase about 5% each in 2021 and 2022, that percent growth is measured from the very low starts in 2020. Nonresidential Buildings starts in 2020 dropped 24%. We need 31% growth to get back to the Feb 2020 high. That will take several years.
With the onset of reduced starts in April 2020, spending began to fall, but only a few percent. The cumulative impact to spending of all reduced starts will be months later than the initial impact. Cash flow shows maximum impact is ~50% to 60% out in time of each individual schedule. The spending in any given month includes input from starts in 20 different months. It’s when a month lines up with all the inputs from reduced starts months that spending reaches its lowest.
In April 2020, we had loses from only one month of new starts which were down 39%. The cash flow schedule declines in the 1st month indicated that spending would drop 0.9% that month. The spending decreases from the next six months of losses in new construction starts indicated spending would drop approximately an additional 1% per month for six months; so down 1% after one month, down 2% after two months, 3% after 3 months, etc.
As expected, actual spending did not follow the pattern set forth by loss of new starts only. In the 1st month, spending actually dropped 3.9%, the difference that month being 3% between what was predicted from lost starts vs actual spending. Nonresidential Buildings Spending actually declined by approx. 4%/month for the 1st four months of the Pandemic, while the losses expected and identified from lost starts increased from 1% to 2%, 3% and 4%. With spending declining at a rate greater than loss of starts would indicate, we have some information to associate with the other cause of decline, delayed projects.
This greater negative performance eventually reached a balance point when actual losses equaled that predicted by lost starts. That would be expected if projects that were temporarily halted were restarted. And just beyond the balance point actual spending, in this case forecast spending, declines are not as great as would be predicted by lost starts. This would occur as the remaining schedule to finish halted projects added some spending to future months that was not in the expected cash flow schedule.
So, to recap, it’s easy to show the cash flow schedule predicting spending reductions caused by loss of starts in the schedule. Also, projects put on hold would show excess spending declines, not otherwise predictable, in the early months and would show unexpected spending increases later.
That’s exactly what the above model shows.
The subtotal line titled MONTHLY SPENDING of $ is construction spending per month, the sum of the contributions from the cash flow of all the still ongoing projects. That shows when greatest impact from lost starts occurs. The low point in spending can be measured in months from the initial event, April 2020. But the combined effect extends well beyond the initial event (reduced starts) which started in April but so far have lasted 10 months. This is why maximum impact of reduced spending for nonresidential buildings stretches over a long period in 2021-2022.
The bottom row shows the difference between Actual spending and predicted spending from starts. That difference behaves exactly as would be expected from projects that stopped spending and then resumed spending later.
The percent spending losses from on-hold projects amounted to no more than a 3%/month loss falling to a 1% loss in spending after six months. From month 10 through month 15, Jan 2021 to June 2021, spending increases 3%/month to 5%/month due to delayed projects resuming spending and completion later than originally scheduled.
So, why don’t we see spending increases from the completion of any delayed/resumed projects?
The spending increases due to resuming delayed projects is far less than the reduction in spending from loss of starts in 2020. By the 10 month, Jan 2021, spending declines attributed to lost starts in 2020 measured 11% decline compared to Feb 2020. By the 15th month, Jun 2021, spending will be down 17% due to lost starts. Delays were never down/up more than 3%/5% in a month.
The magnitude of spending declines from loss of starts in 2020 is three to four times the magnitude of losses, then gains, due to shifting of spending due to delays. The maximum (diminishing) negative impact of delays lasted six or seven months and fell from 3% to 0%. The positive (resumed spending) impact also last for six or seven months. The impact from lost starts reaches a maximum at a point approximately 10 to 12 months (to project scheduled midpoint) after the start. As long as starts are down from Feb 2020, the deepest impact will be 10 to 12 months beyond the last month of reduced starts. Starts in Dec and Jan are still down 20% from Feb 2020. Lower new starts in 2020 cause severe negative impact to spending in 2021 that may reach a maximum impact from May to Dec 2021. For each month that starts continue to come in substantially lower than Feb 2020, that will extend the end of maximum negative impact a month beyond Dec 2021.
Declining spending does not support jobs growth.