If a conversation about startups goes on long enough, it will eventually land on the topic of job creation. To some, startups are the employment savior–the ultimate catalyst for new jobs. Others see startups as a false promise of economic growth.

It’s a divisive issue, and one that grows especially more heated in times of economic distress as communities scramble to find a stimulus to get them back on track. But, these discussions are fierce for a good reason, as rate of employment is an oft-used measurement of the overall health of a company.

Much research has been done on the issue and it tends to run the gamut. Many would agree that, at a basic level startups do, indeed, create jobs. However, the main concern is how healthy these new roles are, and at what expense do they come to the market. Let’s take a look at some of the major ideas surrounding this issue.

For starters, it can be difficult to define “startup.” After all, doesn’t every company start out as a startup? While that’s technically true, there are distinct differences drawn between a startup and a small business, usually around the business model and type of funding used to build the company. Generally, when someone refers to a startup in this context, they are often speaking of a venture-capital backed company, often with a heavy focus on technology.

Taking a high-level view, venture-backed companies seem to have had a massive, positive impact on the US economy. Recently published research by University of British Columbia’s Will Gornall and Stanford’s Ilya Strebulaev shows that now-public companies that began with VC backing are a huge driver of economic growth and employment.

And, the number of unicorns–companies with a valuation over $1 billion–keeps growing. While this does drum up some fears of another tech bubble, it also shows that these companies are able to hold their own.

The success of these giants cannot be understated, but it is worth noting the landscape for building a tech company has changed. Trends like open source, and new cloud options have lowered the barrier to entry. Simply put, tech companies require fewer people to build their products.

“Creating an application in 1999 required hiring large development teams, paying expensive licensing fees, and springing for costly hosting services,” said Eric Jackson, CEO and cofounder of secure enterprise sharing company CapLinked. “Now, apps can be built and hosted in a faster, cheaper manner.”

At the same time, Jackson pointed out, there are new distribution channels like mobile that are creating even more opportunities for entrepreneurs to build a company.

“There’s a vast infrastructure of digital and mobile technologies that tie us all together today that entrepreneurs can build on,” said Larry Augustin, CEO SugarCRM. “It sets the stage for even more accelerated innovation.”

Additionally, as startups create new tools for getting work done, and show the enterprise new ways of doing things. As a side effect, they create new jobs in terms of existing companies needing new employees to help implement tools or change their workflow. For example, a corporation might need to hire someone to help customize the new CRM they recently deployed.

In no other field is this as prevalent as software engineering, where prospective candidates often receive recruitment emails multiple times per day. Hired.com is a company that curates the stream of talent for tech companies. As such, they have a finger on the pulse of the industry.

Courtney Montpas, is the regional general manager of the West Coast at Hired. She said their marketplace of applicants is tripling every year. Additionally, she said they’re also seeing growth on the product side of things.

“After these products or services are brought to market, there is a growing group of people working behind the scenes to keep the customer and client happy with their experience,” Montpas said.

Keep in mind, though, that this job growth comes at a price.

Consider disruption–the calling card of tech startups everywhere. By disrupting an industry, startups change the way the public thinks about the main product or service that industry provides. By changing the way things get done, these startups are creating new jobs, but they are also destroying some in the process.

But, that destruction can be good. Research conducted by the Kauffman Foundation seems to point to the idea that existing companies (not all, but some) are actually jobs destroyers. So, some believe that getting them out of the way is another way of revitalizing the economy. Jackson called it “creative destruction.”

“Blockbuster’s stores are gone. But Netflix, Amazon, Hulu, Apple and many more companies are streaming video in their place,” Augustin said. “Even taking disruption into account, recent studies indicate that startups have a net positive economic effect.”

Some experts disagree, though. Research conducted by Stanford’s George Foster indicates that, around year five of a company’s life, it’s destroying more jobs than it creates. He argues for politicians and pundits to forgo looking at gross job creation, instead looking at net job creation of startups, which factors in job loss.

There is another kind of disruption, too. Artificial intelligence and machine learning pose real threats to automating certain jobs, and startups are leading the charge in this space. According to the Oxford Martin School and Faculty of Philosophy in the UK, 47% of US employment is at a high risk of being automated in the future.

Of course, the argument has been made that automation will simply eliminate certain repetitive tasks and workers will move on to better jobs. Subsequently, a counter argument has been made that those jobs will only be open to those who are highly-skilled or highly-educated.

“US Census Bureau data show that amongst young firms, defined as those less than five years old, job growth is increasing,” Princeton’s Swati Bhatt said. “Since startups are typically less than five years old, this suggests that job creation is supported by startups.”

So, startups seem to be creating jobs, and the Census data seems to back up that idea. But, if you look deeper, that might not be the case. And here’s why:

“After [market] entry, their mortality is very high, about 50% in the first 4-5 years,” said Yale’s Giuseppe Moscarini. Adding, “young firms appear to grow fast in their first few years in part because many of them die and disappear from the data and the dataset, so the survivors grow fast, but it is to a large extent survivorship bias.”

Also, Moscarini said, that data is compared against existing firms, which tend to destroy jobs on net. Still, the mortality rate is alarming. Factor in pivots in the business, or running out of capital and you can see more jobs lost.

It would be wrong to look at this data divorced from the changes in the workforce as well. Consider the five year number that keeps coming up. Maybe five years is long enough to be considered a “lasting” job among today’s workers. According to the Bureau of Labor Statistics, the median number of years that employees are spending with an employer is 4.6. So, maybe we are just seeing a bigger overall shift in how people work. Although, some are disputing the BLS data as well, saying that it’s too difficult to measure along generational lines and changing jobs is better for the worker.

The question of whether or not startups are creating jobs is far too complicated and multi-faceted to answer with a simple yes or no. Entrepreneurship and innovation continue to be a major part of the US economy, yet companies like Google and Amazon don’t even make it on the top 50 list of largest US employers. Still, the work they are doing is touching the lives of almost every citizen in the US. All in all, it is an intricate relationship between tech and business and it is getting more and more complex every day.

If startups had to describe their relationship with job creation on Facebook, it would most definitely read: “It’s complicated.”