Ratios, benchmarks, substitutes, voodoo and a pinch of luck
It is hard to believe that it has been two years since I was faced with forecasting market growth rates for WMS and Warehouse Automation amid the COVID-19 pandemic. At the time, I was writing about the COVID pandemic and how similar events are occurring that increase market uncertainty. These events make accurate predictions very difficult. Well, here we are just two years later in another highly unusual market context. I had Logistics Viewpoints readers in mind as I attempted to incorporate the new factors into my WMS market forecast. It has struck me that many of you also have the responsibility of navigating this dynamic environment of post-pandemic demand shifts, inflationary pressures, supply constraints, rising interest rates and a possible economic downturn. I’ll take this opportunity to discuss some of the approaches I use, hoping that some of my content will spark a valuable idea to improve your approach.
New factors with limited history
I tend to use time series analysis as an anchor for my forecast, as I suspect many of you do. It only makes sense to use the past as input for future expectations. However, time series only provide general guidance and not well-informed expectations. I once heard a prominent economist refer to long-term drivers as “anchors” for expectations. I consider time series as “anchors” for my forecasts. Not entirely coincidentally, the link above points to an interview in which the same economist points to the Federal Reserve’s liquidity injection as an anchor of expectations. The highest and most persistent inflation since the early 1980s is leading to lower liquidity and higher interest rates – and added uncertainty in forecasting economic activity. To take a step back, the current inflationary environment is partly due to supply shortages, which in turn are partly due to pandemic-related restrictions. These current factors, in addition to the recent atypical activity (whipsaw demand patterns), make time-series trends less relevant than in more stable environments. So what methods can we use to better estimate future trends?
Qualitative review of the explanatory variables
A Examination of the explanatory variables can influence expectations, forecasts and measures. However, the relationship between variables can shift, making them less reliable indicators in certain circumstances. For example, in a recent CNBC interview, Ben Bernanke noted that the Federal Reserve likely examined the unemployment rate and total employment in early 2021 and concluded that there was plenty of room in the labor market. With hindsight, however, it became clear that pandemic-related factors influenced the relationship between these employment statistics and labor slack. In fact, the pandemic created a distortion that led to fewer labor shortages and greater inflationary pressures.
In my forecasting process, I wonder if there is reason to believe that independent variables responsible for influencing my growth expectations have changed. I’ve found this to be helpful in “tweaking” likely growth within markets. For example, interest rate hikes tend to deter lower priority investments and those that require leverage. My colleagues and I have used past relationships between GDP and capital spending to estimate the potential magnitude of changes in capital spending in a recessionary environment. A review of similar comparisons can be instructive in an environment of rising interest rates. Inflation has prompted me to reconsider the likely magnitude of the importance of certain purchases, as employee salaries are not increasing at the same rate as inflation and purchasing power is falling. As a result, spending on non-durable goods is likely to decrease and there should be a shift towards substitutes due to price differentials. Substitutes are also relevant due to today’s supply bottlenecks. A review of likely substitutes can provide information about potential changes in demand for particular categories of items.
Looking back at earlier significant events
When data on causal factors are not readily available, it can be instructive to examine the behavior of specific industries or economic activities in response to disruptive events. In the first half of 2020, my European colleagues Florian Güldner and David Humphrey conducted a thorough scenario analysis to assess the potential impact of the coronavirus on automation markets and supply chains. Their model used historical data on capital expenditures by publicly traded companies, broken down by end-user industry, along with the impact of past regional economic shocks such as the Tokyo earthquake, Hurricane Katrina, the SARS virus outbreak, and other events. Their analysis showed which industries tend to experience large swings in capital spending after an economic shock. Of course, analysis of this type uses different events with different mechanisms of influence on activity, but it can still provide some useful insights that can guide further analysis.
As my statistics professor said, “Don’t worry about getting right or wrong with forecasts. I’ll spare you the suspense, you’ll be wrong. What matters is how close you get to correcting your forecast.” Hopefully, something in this article will spark a useful thought for you and improve your forecast so you can… get a little close.