The Architectural Billings Index (ABI) is an instrument used to forecast non-residential construction spending in America 9-12 months into the future. Every month, the American Institute of Architects sends out their "Work-on-the-Boards" survey to about 700 architecture firms, which survey asks them to report on billings growth or contraction as compared to the previous month. If billings were shown to increase, then there is a high likelihood that construction spending will increase as the buildings that the architects design from those billings is eventually built. If billings increase for several months sequentially, the future earnings of businesses whose products and services are used in the construction of non-residential buildings will likely rise as demand goes up, which has natural implications for the common stock of those companies.
The index is used by many firms to predict overall economic activity and can, therefore, be used as a gauge for whether or not to start or stop certain projects. Johnson Controls (JCI) uses the ABI to track the business cycle and its impact on staffing requirements. AAON, Inc. (AAON) routinely references the ABI in their conference calls to inform investors of future business conditions and demand for their HVAC products. ADD Inc., a design firm, uses it to prepare for fluctuating market conditions. Gilbane, Inc., a large privately held real estate development firm, uses the ABI to make decisions about multi-year projects and to manage its overall budget. Clearly, a reputable and oft-cited resource, my intent today is to get into the details of the index and discuss both its strengths and limitations when used to make investment decisions.
The ABI is a diffusion index, "a method of summarizing the common tendency of a group of statistical series." As participating firms report on their billings, the percentage of those who report an increase in billings is added to half of the percentage of firms who reported no change. The diffusion index is centered around a value of 50: anything above fifty means billings went up in aggregate and any value below fifty means billings went down. If 45% of firms report an increase and 20% of firms reported no change, the ABI score would be 55, indicating a general upward trend in billings. Firms report an increase if billings went up 5% or more, a decrease if billings went down by 5% or more, and anything in between is to be reported as no significant change.
The ABI composite number is the average of four separate reporting regions:
*Image from October 2018 ABI report
This has natural implications for publicly traded companies that are regionally based. For the trailing 12 months, the northeast has seen mostly decreases in billings. However, the four regions seldom move in tandem, and trend lines can move sharply. Last year, the northeast was strongest amidst the four for five straight months. Trying to tease durable investing strategies out of the regional data is more tricky than using the composite number, which will be discussed later.
This methodology has considerable limitations. First, results are not adjusted for the size of the firm itself or the absolute value of the billings they make, only the direction of the amount. That means that a small firm who only handles a couple million of billings gets the same weight in the index as a firm that handles hundreds of millions. Forecasting construction spending is questionable under these premises, where relative values from one month to the next can't translate reliably into a measurement of absolute dollars to be spent in the economy.
Second, the index makes no adjustment for factors that may influence billings that have nothing to do with the actual economy. For example, a firm may develop a bad reputation, and a reduction in billings from month to month may simply mean that they aren't staying competitive. The converse can also be true; patrons may flock to a firm that has a great reputation, and the sequential increase in billings from month to month could even be happening in a slowing economy.
Third, there are no controls in place to enforce the accuracy of respondents. An email link is simply sent to each firm each month. We don't know if the responses are from the director of each firm, a secretary, or someone simply delegated the task. We don't know what efforts are made (or lack thereof) by each firm to ensure accuracy, or if ballpark figures are merely used. "Hey Jim, do you think billings went up or down this month compared to last" is hardly scientific, but in the absence of any incentives for each firm to respond with exactness, taking the time to complete the survey may, in fact, be a nuisance and something just done quickly. Basing economic forecasts on perhaps primitive computations is faulty.
Fourth, some of the billings activity at the reporting firms may have little or even nothing to do with things that will result in construction activity. For example:
While the sample size is substantial enough (700 participating firms) to mitigate the effects of some of these disadvantages, it should nonetheless be kept in mind that this data is nuanced, there is considerable margin for error, and the ABI has been wrong in the past. According to a paper on the ABI web page:
Coming out of the 2001-2003 nonresidential construction recession, the ABI recovered faster than did construction activity, but it turned out to be a false recovery, as the ABI was above 50 for three months and then fell below 50 for 15 months."
For those using the ABI to inform their decisions about when to get in or out of the market, or when to get into or out of specific businesses, these errors should be kept in mind.
One thing that the ABI does relatively well is predict overall construction spending, which can reliably translate into overall economic health and subsequent movements in stock market indexes, but with considerable limitations regarding the strength of certain movements, particular to the upside. Researchers at the AIA have published papers that show the correlation between the ABI and actual construction spending and have found that the ABI is an accurate indicator of spending with 11 months' worth of lead time. The following graph illustrates this point:
*Images from ABI 2014 White Paper
Now compare these graphs with charts of several market indices, the S&P 500 (SPY), The Russell 2000 (IWM), and the Dow (DIA) over the same time frame:
There is an obvious and visible connection here. The main limitation of the index is its inability to predict the strength of upwards movements. But considering the fact that beating the market has at least as much to do with protecting the downside as it has to do with capturing the upside, the ABI can help avoid catastrophic loss as it reliably leads economic downturns if the index severely dips below 50. Let's zoom in on those bear market periods to get a more detailed view.
The ABI plunged below 50 between July of 2000 and January of 2001, forecasting the ensuing tech wreck. About six months later in June of 2001, percentage change in construction spending as compared to the prior year dipped into negative territory. Had you heeded either of these indicators and got out of the market, you would have avoided the considerable drop in equity markets that followed. Obviously, listening to the ABI drop that came first would have been a near perfect market timing, getting out near the top of the prior bull market.
Timing the recovery would have been a little more challenging. The ABI ticked above 50 for the first time since the plunge in April of 2002. But it quickly went back down again until perking back up around the end of 2003, which almost exactly coincided with the recovery in construction spending. Had you gotten back in the market when both the ABI and construction spending turned positive, you would have enjoyed respectable returns, albeit missing the actual bottom in 2003.
And what of the financial crisis.? The ABI sunk severely below 50 around January of 2008. Construction spending dipped negative around November of the same year. The behavior of the stock market in that same time frame behaved thusly:
Had you heeded the ABI, you would have gotten out (again) near the top of the cycle. By the time construction spending followed, a lot of damage had already been done, but you still would have saved some money had you gotten out when it dipped into negative growth.
The ABI went above fifty in November of 2010. Similar to the previous instance, it went slightly negative again shortly thereafter and stayed just below for a time before jumping back above. Construction spending went positive in September of 2011. By this time, the recovery was well underway.
In this instance too, the ABI was a phenomenal indicator of when to get out of the market. Trying to time the recovery with the ABI, however, was much harder, missing the strong initial leg out up out of the March bottom. In summation, the ABI predicts well a drop in construction spending, which indicates a slowing economy, that results in a suffering stock market.
At this time of considerable market turmoil, an eye on the ABI can provide information that will allow for appropriate asset allocation and portfolio positioning. The tariff situation is being felt across various industries, people continue to worry about the yield curve, and what happens to earnings AFTER the prop from the tax cuts and jobs act may spell trouble for markets ahead. But what about the ABI? Is it up or down? The following graphs show data from January of 2017 - October 2018:
*Image from January 2018 ABI report
*Image from October 2018 ABI report
It is worth noting that since the financial crisis the ABI has been pretty range bound between 50 and 55, whereas in prior years it wasn't uncommon for the index to flirt with 60. As can be seen from the first set of graphs, the index moved drastically below 50 preceding both of the past two recessions. For those using the ABI to try and time the market, I would say wait for a sharp move below 50. A slight dip may mean nothing. If the ABI comes out this month and reads 46 or below, I would take caution.
Keep in mind that the ABI has only been around since 1995 and has, therefore, only been through two recessions. A data set of two is not nearly enough to claim that what has happened the last two times will happen the same way next time and people should pivot accordingly. But it is useful at best, and interesting at least.
Something to keep an eye on in the future is how the AIA is looking for a reliable way to forecast billings, and therefore generate even more lead time for construction spending:
In addition to collecting data on architecture firm billings from our ABI survey panel every month, panelists are also asked about inquiries for new work (e.g., bids, requests for proposals [RFPs], solicitations, invitations for interviews). As with billings, they are asked to indicate whether those inquiries have increased, decreased, or remained the same from the previous month. However, because reporting trends in new project inquiries is rather subjective on the part of the panelists, the AIA has been investigating a more rigorous way of estimating future levels of billings activity.
Responses by ABI panelists indicated that, in their view, the most accurate predictor of future design workloads would be the monthly change in the dollar value of new design contracts-agreements between the client and architecture firm on the scope of and compensation for the design activity. In a recent AIA survey, architecture firms indicated that 88 percent of their projects have a signed design contract, and that nearly all billable work for a project is typically completed after that contract is signed. However, in an effort to include projects with more informal arrangements, firms are asked to include all new projects in which they have a letter of agreement or a formal authorization to proceed-effectively anything that provides assurance that a firm can bill a client for future design services.
The AIA began collecting data on design contracts in October 2010 and now has three years of data-enough to seasonally adjust the index, although it remains too early to determine the exact relationship between design contracts and subsequent architecture billings. However, through visual analysis it does appear that when firms report an increase in design contracts, an increase in firm billings follows within the next six months or so. In the future, this new indicator should prove valuable by providing even earlier insight into future design and construction work.
I am not a market timer. I don't see myself EVER pulling all my money out of the market. The market moves for reasons outside of yield curves, tariffs, or the ABI. However, tools like these can be used to cement opinions about when to take some profits off the table and/or move into different asset classes or sectors. What I like the most about the ABI is that it isn't based on what has happened in the past or trying to chart a trend. The ABI is all about the future. What businesses spend on architectural consultations and billings is a natural forecast of what those same businesses will eventually spend on buildings, offices, industrial plants, etc., which has obvious implications for the production of goods and services. The construction of those buildings has obvious ramifications for construction supply companies. Among the many tools that get oft-cited as portending market movements, the ABI is rarely referenced yet is, in my opinion, one of the better economic harbingers. Its limitations are real, but its advantages should be part of each investors tool-kit.
This article was written by
Disclosure: I am/we are long AAON. I wrote this article myself, and it expresses my own opinions. I am not receiving compensation for it (other than from Seeking Alpha). I have no business relationship with any company whose stock is mentioned in this article.