The Gathering Tech Economy Storm

Recently, I've begun seeing signs that I've been expecting for a while, even if I've not been wanting to. There is a tech recession coming. Like Hurricane Harvey, it has been sitting off the coast for a while as a tropical storm, but it has the potential to turn into a category 4 hurricane all too easily.

Oscillating Recessions

I started working in IT back in the early 1980s. After thirty five years, I have enough data points to at least hint at what's approaching. There was a recession in the 1984-6 time frame, another in 1991, a third in 2000, and a fourth in 2009. You can also go back almost a century and find a similar set of patterns, with significant recessions about once every 8-10 years, and "slowdowns" every 4-5 years. What's worth noting as well is that there's also a super-pattern, with equity oriented recessions occurring about once every 17 years (every other recession), and real-estate (usually housing) recessions then being the counterpart. Now, the stock market has a tendency to fall dramatically in both, but in the 2009 recession, the recovery in the stock market was remarkably fast, even though the effects due to so much housing-related money evaporating have lingered through most of the resulting recovery.

What is significant about the 2000 recession was that it was the first time that technology existed as a distinct sector in the economy. When the market crashed, for most of the country the impacts were comparatively minimal - many people were inconvenienced, but they could have been excused for thinking that the economy wasn't really all that bad. If you were in IT, however, it was nuclear. It was Game of Thrones level blue zombies amassing on the Northern Wall. Programmers who were living in mansions paid for by stock IPOs suddenly found themselves living in their parents' basements, or underneath overpasses, with Will Code for Food becoming the central mantra.

In 2009, on the other hand, things slowed down for IT workers, but they were among the first to get jobs even as the economy struggled to deliver. It hit the Heartland hard, however - pretty much spelling the end of traditional manufacturing in many parts of the Midwest and Southeast. So it's worth understanding why these recessions were so different.

The IT Boom and Bust Cycle

IT is not a single sector. It is rather the application of algorithms, networking and fast computing power to other sectors. Most companies do not start projects because it's cool, despite what may be the impression that programmers have. They start IT projects to gain an edge in a competitive environment, whether that be offering new modalities of interacting with the companies products or services, better tools for obtaining useful and meaningful analytics from every potential data stream, tools to optimize the production of goods and the application of services, or finding ways to decrease the friction in payment channels. In short, for every IT project that is begun, there has to be a distinct business need to be met first.

Right now, a lot of traditional businesses are facing significant headwinds. Brick and mortar retail outlet sales are shrinking dramatically. As B&M slows, fewer goods are ordered, fewer miles are traveled by trucks, fewer package containers are sent on ships. There is less money to be invested in R&D, especially for investments that have shown comparatively little return on investment. Big Data was a marketing ploy that for a while looked like it might succeed, but the last decade is littered with the remains of big data projects that failed, while the number of success stories are discouraging in their paucity. Now credit is beginning to tighten again, money that was essentially available for free is becoming expensive again.

A software project has a distinct life-cycle. Usually there is a six month period that is purely business development - determining needs, talking with vendors, establishing a schedule. This typically involves business analysts trying to determine if there is sufficient potential for ROI to invest. When credit is tight, many of these projects go into limbo and are never green-lighted.

As budget is found, that's when IT begins to come into play. The first IT players are typically consultants, enterprise architects, scrum-masters and designers. These are the people who are going to turn the business requirements into specific breakdowns for developers, UX designers and others. Their role is always highest at the beginning of projects, but they cede power and authority as the project moves towards completion. They are also traditionally the most expensive and most senior.

Next hired are technical managers, senior programmers, database admins, data modelers and DevOps. They are the ones that insure that the infrastructure is in place, and usually stay for the bulk of the project. Once these roles are established, you get developers in seats, each tasked with solving certain component problems, then moving onto other components when their task is done.

After the devs, you need to ramp up the testers, the ones who do quality assurance, check for bugs and write the documentation. These are typically junior developers, under the supervision of senior devs and project managers. They are also quite frequently the ones that will go on to be the maintainers once the project is done and shipped.

From start to end, this process takes roughly two years, regardless of whether the project uses waterfall methodologies or agile. As each six month phase ends, you see a certain degree of attrition from the previous phase - architects get moved onto other projects, senior developers find themselves pushed into management positions, some junior devs gain seniority, until by the end, the bulk of the devs are released to move onto new projects, either with that company or elsewhere.

Once a project is done, if it was successful, it will satisfy a baseline of needs that remove that project from the docket, at least for a decade. Every subsequent project thus takes longer and longer to green-light, because it had lower priorities. If the project was unsuccessful, there will also be a wariness towards investing in new projects, especially with untried technologies, and that too will impact what's in production. This means that realistically, for any given company, there will likely be a maximum of three major projects before the window closes on R&D as credit starts to become dear again.

Ironically, this late IT cycle can masquerade as full time employment, or even over-employment. Remember that not all companies start at the same time. The first cycle is typically smaller, the second larger, and the third larger still, so that the total number of people in demand can jump dramatically as successive waves overlap. Yet one of the first signs that a business cycle is ending is that your lead people - your enterprise architects, business analysts and modelers - find it harder to find new gigs. Consultants usually see this first, but full time employees may find themselves encouraged to take early severance in order to get their very expensive salaries off the payrolls, because senior managers are seeing the numbers and knowing they are not going to be needed.

Storm Clouds

Around November, 2016, I started hearing that people who should have been fully employed - enterprise architects, data modelers, business analysts - were having trouble finding work. Contracts that were in the works were being renegotiated or put on hold. Consulting companies that had banner years in 2014 and 15 were seeing YOY numbers stall or drop in 2016.. These positions are expensive to find and fill (and frankly to fire), so watching these people getting pink slips suggested strongly to me that we are near the end of the business cycle.

This has been borne out by successive database vendors announcing layoffs in both their sales forces and senior engineering staff, especially in the NoSQL sector which has seen significant growth previously. Most have been under the radar, but the decision to buy or install databases is one of the first that IT projects need to make, so when such vendors begin to trim both sales and tech staff, it means that they haven't been hitting their projected growth numbers. Sometimes companies miss, but when it occurs across the industry, then this is an indication of global forces - people are not starting as many projects that they need ANY database for.

Another sign that we're going into a slowdown has been the anomalous rise of programmer cooperatives. Certain skillsets are typically in great demand because there are relatively few suppliers - C++ developers (especially with Objective C type background for IOS), Java developers with Hadoop, Node devs, just to name a few. Yet I am seeing now about once a week emails or Linked In posts where "groups" of like-talented devs are looking for projects, and even in my circle (which are primarily early tier talent), we have support networks where we pass on potential opportunities that are simply not feasible to meet, but even there, I'm seeing fewer and fewer of those opportunities showing up to pass on (or take, for that matter).

So why are there so many job ads out there - surely the job market is more robust than I'm painting it? First, it's worth going back to the cycle model. Most companies, having committed to a project, will not terminate it unless it has completely gone off the rails. If you assume that the slowdown hit early tier IT folk first, there is still a significant chunk where the decision to staff has been made, where the developers are working. Most of these will stay on until the end of the project. If the slowdown started in Nov 2016, this suggests that we will start seeing unemployment to start creeping back upwards by November 2017, as programmers laid off of projects find that there are fewer jobs available.

Bubbles Bursting

For recruiters, I think 1Q2018 will be the watershed year - they will have more people looking for fewer jobs, and this means lower commissions as wages begin to drop. Given the margins for recruiting are already becoming paper-thin, I suspect that recruiting, ironically, is going to be hit especially hard this time.

What about the startup space? It's a different dynamic, but is still a predictable one. We're swimming in free and near free software, services and apps. The software services sector is saturated, there are so many NoSQL databases out there now that you need a couple just to keep track. Data scientist salaries are falling from nosebleed territory even as software replaces the 90% of all functions that a data science typically performed. You can't throw a bitcoin without hitting a blockchain somewhere.

Yet there was something I noticed after the 2000 recession that I think holds true today. A recession is a test. It tests whether a company has what it takes to hold together when times are tough, when funding is scarce, when the foosball tables and gourmet delivery gets replaced with stale pizza and 60 hour weeks. I think a lot of startups will fail that test, with investors out their money. At the same time, that period also is the "s**t just became real" time, when ideas are quite-literally stress tested. It is perhaps not surprising that so many of the biggest names in software got their start in or just before a recession.

There's a lot of things that people in IT can do now, to help alleviate the worst pains of a recession. Start paring back on unnecessary expenses. If you have software that you're creating on the side, make sure it's in a maintainable state. Pick up writing gigs, teaching, curriculum development. Look at options for going back to school. Perhaps most importantly, lay the groundwork for potentially extended periods of unemployment. To the extent possible, get your credit rating in good shape, because it'll take a beating. If you own stock, take a defensive position, and think about your mix.

The Good Side of Bad Storms

Recessions can be hard, but they are a necessary part of the economic cycle. They usually reduce economic imbalances. They also give a number of technologies that were just getting off the ground, often facing stiff competition for resources and the demand to get it out the door, a couple of years to gestate, to get the kinks out. Autonomous vehicles, for instance, are "nearly there", but there are still issues, both technical and social (regulatory, political) that need to be worked out. Blockchain tech such as Etherium needs a better story for managing the energy costs of calculating tokens, and so forth. Often, after a recession, what emerges, may be far superior to what would have resulted had ideas not had the longer window to gestate.

Finally, recessions can help to winnow out older innovations and social ideas that are no longer effective, and exist only because "well, that's the way we've always done it", though again, I think there is a distinction between equity and bond (mortgage) recessions. Housing recessions most heavily impact those places where housing was purchased under exotic instruments, and usually hits mortgage banks hardest. The coming recession will more likely hit bond financed projects - malls, big box stores, retail in general - hardest. While the damage from Hurricane Harvey in Houston will send gas prices up (which exacerbated the crisis in 2008) the fundamentals of oil production are still dictated primarily by global conditions (though people in the Southeast will likely be hit hard by gas price spikes for awhile, potentially pushing them into recession early).

At the same time, I don't believe that the coming recession will likely be as severe as 2009, even in the tech sector, though it may be what finally pushes a lot of brick and mortar retail out of business entirely. Tech is too widely spread out at this stage to be considered a sector in and of itself. It's now health tech, or retail tech, or automotive tech, or some other industry's tech, and as such I think that tech workers will likely be insulated from the worst effects of the recession of 2018.