New construction starts drive construction spending. For all the discussion regarding the monthly rise and fall of spending, most of the spending in any given month is already predetermined since two thirds of all construction spending in the next 12 months comes from projects that were started prior to today. This is commonly referred to as backlog.
The pattern of spending does not follow the pattern of new starts which can fluctuate dramatically. It follows the pattern developed by the cashflow from all previous starts. Data for new construction starts is sourced from Dodge Data & Analytics. Cash flow is developed independently. Here’s a much simplified example of cashflow: a new $20 million project start is to be completed in 20 months, therefore we expect this project to generate $1 million of spending every month for the next 20 months.
This plot is an Index, so the ratios of starts and actual spending show the relative volume of each of these three major sectors as compared to each other.
Nonresidential buildings new construction starts were elevated for 16 out of the last 24 months. Starts were strong from February through July of 2015. A slowdown occurred in the second half of 2015 but the last four months have been gaining slowly. It looks like the backlog of elevated starts will keep spending rising at least until the end of 2016 before we see a slight dip in spending.
75% of all nonresidential building spending in 2016 comes from projects that were started between early 2014 and the end of 2015. Each month, new starts generate only 4%-5% of monthly spending. As we start the new year, backlog accounts for 95% of January spending. We know a lot about spending within the next few months, but what we have in backlog for December at the beginning of the year from previous starts accounts for only 50% of December activity. We will add about 4-5% more to December backlog from new starts each month this year.
Five out of six times in the last 18 months that nonbuilding infrastructure new construction starts jumped 25% to 50% above the running average it was due to massive new starts in the power sector. Some of these projects are worth several billions of dollars. While this causes new starts to fluctuate wildly, these projects sometimes take four to five years from beginning to completion, so the cash flow is spread out over a very long period, therefore spending does not experience the same magnitude of monthly change as starts.
80% of all nonbuilding spending in 2016 comes from projects that started from mid-2013 through the end of 2015. New starts each month generate only about 3% of monthly spending.
The average of residential starts for the last three months is higher than any time since 2007 when residential starts were already on the decline by 24% from the previous year. The volume of residential starts predicts that spending should be higher than it is currently. This could mean that some starts have been delayed. Or, it could be because residential starts have the shortest duration, they may be the most difficult to predict spending from starts.
55% of all residential building spending in 2016 comes from projects that started between late 2014 and the end of 2015. New starts each month generate almost 10% of monthly spending.
(6-5-16) RE: a discussion related to a decline in nonresidential permits suggests nonresidential spending will decline. Yes, but at what rate? Permits are directly related to new construction starts. Since every month of new starts has an impact of only 4-5% on nonres spending in every following month for the next 20-25 months, then a 10% drop in permits in a single month would cause only a 0.4% to 0.5% reduction in spending in each of the following 20-25 months. It would take a prolonged trend of declining permits and therefore declining new starts to really see a dramatic decline in spending, and then the greatest effect would be well out into the future.
Construction volume is not the same as construction spending.
Spending is the number nearly everyone follows. Volume is spending minus inflation.
Two years that show the greatest differences between spending and volume highlight the affect of inflation. In 2004 and 2005 total construction spending grew by 11% and 12.5%, but after inflation, volume grew by only 3% and 2%. In the most recent year, construction spending in 2015 grew by 10.5% but total construction volume grew by only 7.5%.
For the four years 2012 through 2015 construction spending grew by 35% but after inflation volume grew by 21%. Inflation accounts for 14% of spending growth.
Annual construction inflation varies for residential, nonresidential buildings and nonresidential infrastructure, and it varies sometimes so widely that each should be used only to adjust that specific group. Since 1993, long-term annual construction inflation for buildings has been 3.5% per year, even when including the recessionary period 2007-2011. During rapid growth periods, inflation averages more than 8% per year.
Historical average volume growth over the last 22 years is grossly distorted by the recession. Volume declined in 8 of those 22 years. In the worst three years of the recession, 2008, 2009 and 2010, volume declined by 28%. If we take out those three years the typical growth period averages are more apparent. The historical average volume growth in construction with recession data removed and after adjusting for inflation is 2% per year for 19 years.
Adjusting for inflation is changing current dollars to constant dollars.
Current dollars = dollars are reported in the value of the year reported, 2008 = 2008$, 2015 = 2015$. News reports almost always refer to current dollars and therefore do not account for the influence of inflation.
Constant dollars = all dollars adjusted to represent dollars in the year of comparison. This adjusts for inflation so 2008$ (and all other years) are converted to equivalent 2015$.
It’s not too hard to understand why we need to look at constant dollars when you think of it in terms of building a house. For example, a 2,500 sf house built in 2001 may have cost $250,000 then to build. Today to build that exact same house may cost $400,000. The house is no different, so volume remains the same. The only thing that changed is inflation. With respect to constant dollars for the same volume, $250,000 in 2001 dollars would be equal to $400,000 in 2015 dollars.
The common comparison is to look at growth in construction spending from year to year. What that does not tell us is how much of the spending growth is inflation and how much is a real increase in construction volume.
Constant dollars makes a huge difference in the analysis. Adjusting all previous years of spending allows us to compare changes in volume growth from year to year. This plot of total construction dollars shows current dollars would indicate we are now only 7% below the previous high and we’ve had growth of 37% from the recession low. Constant dollars, adjusting for inflation, shows volume is still 17% below the previous cycle high and we’ve had growth of only 22% since the recession lows.