Peter Cappelli: HR management practices companies refuse to fix

The relationship between what HR management research says and what organisations actually do has never been straightforward. Decades of evidence on hiring, training, talent development and workforce planning sit largely unread by the practitioners who would benefit most from it, while a vast industry of consultants and vendors shapes HR decisions in ways that have little to do with what the research shows.

That’s the perspective of Peter Cappelli, George W. Taylor Professor of Management at The Wharton School and Director of Wharton's Centre for Human Resources, one of the most cited management scholars in the world. His work has consistently challenged the perceived wisdom of the HR industry, from the myth of the skills shortage to the persistence of generational stereotypes. In a conversation with UNSW Business School’s Professor Karin Sanders, he traced the gap between what research says and what organisations actually do, and explained why the gap is not closing.

Prof. Sanders: Many popular HR narratives, such as skills shortages and generational differences in the workforce, remain influential despite limited evidence. Why do they persist?

Prof. Cappelli: There is an explanation that might surprise people. My former student and current colleague, Shoshana Schwartz, and I totalled up the size of the industry that supports human resources. Many HR tasks have been outsourced over the years, some offshored. But the total size of that industry is roughly US$1.6 trillion, which is comparable in scale to the gross national product of Italy. These firms spend around 10% of their budgets on marketing each year. That is approximately US$160 billion being spent on generating and amplifying narratives.

The vendors are not particularly interested in being right if being right does not make money. They have no interest in telling you there is nothing to see here. So they generate problems and then generate solutions for them, and because they run all the conferences and produce all the reports, they control the narrative. If you are an HR manager, you probably receive 10 to 15 messages from vendors every day, each pushing a problem and a potential solution.

Learn more: In digital transformation, managers are ‘mission critical’

The generational one is the most striking example. The US National Academy of Sciences produced a report on this and concluded that it was all unsupported. There is no evidence for generational differences in the workplace in the US. And if you are outside the US, you should largely ignore that literature entirely, because it is US-specific. You might very well have generational differences in China or India, but they would not look anything like the American version. Yet the narrative persists because there is a whole industry of consultants whose business depends on it.

Prof. Sanders: Your 2008 book, Talent on Demand, argued that companies should manage talent more like a supply chain. Reflecting on it now, what has changed?

Prof. Cappelli: Very little, unfortunately. The context for that book was my earlier work on the breakdown of lifetime employment. Once companies were no longer managing talent internally across long careers, the question became: how should you manage it instead? The argument in Talent on Demand was that business had become too uncertain for long-term planning and that companies needed to actively manage that uncertainty rather than simply be surprised by it.

When the book came out, just before the Global Financial Crisis, there was genuine interest in those questions. Then the recession hit, companies fell back into survival mode, and attention drifted. What happened after that? Essentially, nothing changed. Companies still wait until the last moment and then go and hire someone. They offer no or only a few hours of training to their employees. They are doing less and less on development. Their solution to uncertainty is just-in-time hiring, which is not actually a solution; it is a way of living with a problem rather than addressing it.

Subscribe to BusinessThink for the latest research, analysis and insights from UNSW Business School

If your answer to every talent requirement is to hire from outside the moment you need someone, you end up paying a significant premium to poach from competitors, and you churn through your own employees because you are not developing them. That is not optimal. It is just the path of least resistance.

Prof. Sanders: The statistics you cited on promotion rates are striking. Can you explain what is happening there?

Prof. Cappelli: ADP, the payroll processing company, tracks approximately 80 million employees in the United States. According to their data, only 1% of employees receive a promotion in a given year. That is a dramatic departure from what used to happen, when career progression was reasonably predictable because organisations were investing in internal development.

But the more striking figure is what happens after a promotion. Thirty per cent of employees who get promoted quit within a month of receiving that promotion. The reason seems to be that many employers treat a promotion as a reward – a signal that the employee is valued. What they do not appreciate is that a promotion also signals to the outside labour market that this person is more valuable. So the employee gets promoted, receives outside offers, and leaves.

There is a related dynamic here, and it connects to the broader pressure on organisations from financial markets and CFOs responding to those markets. That pressure makes companies skimp on other factors, such as pay, training, and adequate hiring. The cost of that parsimony shows up in turnover, but most organisations are not measuring it carefully enough to connect the two.

Learn more: Amy Edmondson on psychological safety in an uncertain world

Prof. Sanders: How should organisations think about the cost of that approach?

Prof. Cappelli: The simplest indicator of whether an HR function is operating with any seriousness is whether it can tell you the cost of turnover. It is astonishing how few HR professionals know their own organisation's turnover costs. If you do not believe turnover costs are anything significant, you will never take steps to reduce them.

Turnover is expensive in ways that go beyond recruitment fees. You lose institutional knowledge. You lose the investment you have made in even the limited onboarding you provide. You lose productivity during the gap and the ramp-up period. If you are not measuring those costs, you cannot make the case internally for doing things differently.

The same logic applies to training. Employers gradually persuaded themselves over several decades that training was someone else's responsibility, such as the governments or universities. That shift created a situation where employers now expect people to arrive ready to perform on day one.

Prof. Sanders: You have written about the mismatch between education and the labour market. What should universities and employers be doing differently?

Prof. Cappelli: The mismatch sits primarily on the employer side, not the university side. In the 1990s, when we began seriously examining these questions, young people in the US were taking roughly eight years to settle into stable employment. Employers could afford to be selective because the labour supply was large. What they told us at the time was straightforward: give us people with the right behaviours – reliability, responsibility, basic work readiness – and we will train them for the rest.

Learn more: Why AI systems fail without human-machine collaboration

What shifted over time was that employers came to believe they should not have to train anyone. That became the orthodoxy. I wrote a book examining this called Will College Pay Off?, and the conclusion was that what employers actually valued most was work experience, specifically internships and prior roles, not the credential itself.

Universities are not well-positioned to replicate on-the-job learning. You can arrange apprenticeships, but in the US model, at least, you are essentially paying tuition for someone to do work experience they could probably access more cheaply through direct employment. The more fundamental problem is that employers now pay a large premium to hire people with ready-made skills from the outside, rather than developing them internally. They have decided that training is not their job, and they are paying a significant but largely unacknowledged cost for that decision.

Prof. Sanders: You have observed a persistent gap between academic HR research and HR practice in organisations. What drives it?

Prof. Cappelli: Our colleague Sara Rynes did a study some years ago that asked practitioners what they believed the right answer was to a range of questions where the research evidence was reasonably clear. The practitioners were systematically wrong. That pattern has been replicated in the Netherlands, Turkey and elsewhere. It is consistent across contexts.

Part of the explanation in the United States is that it is not unusual for people to reach senior HR positions with no HR background at all. In finance, that would be unthinkable. In marketing or information technology, you would never appoint a chief officer with no domain expertise. In human resources, it happens routinely. We often see lawyers appointed to lead HR functions because of their employee relations background. But employment law is a narrow slice of what HR involves, and expertise in that slice does not transfer to the rest.

Learn more: How algorithmic HR systems impact motivation and wellbeing

Academic HR and organisational behaviour programs at elite universities have been progressively reduced in scale and status. The function itself lost the status that it once had. In the 1950s in the United States, human resources (then closely connected to labour relations) was rated the most prestigious function in business by other executives. That standing has eroded considerably.

Prof. Sanders: There is a related argument about the type of research HR academics produce. What is your view on the gap between the studies that get published and the questions organisations actually need answered?

Prof. Cappelli: It is a real problem. The questions that matter, such as what happens when you introduce AI into a workplace, and what remote work actually does to collaboration and productivity, are being studied in economics and organisational sociology, not in HR or organisational behaviour. Those fields have the data access and the methodological flexibility.

I published a review article this year on artificial intelligence and workforce outcomes. Almost all the credible empirical work on that topic was coming from computer science, not from HR or psychology. Part of the reason is that psychology and organisational behaviour operate within a conservative paradigm that requires new work to anchor itself in existing theory. If you are studying a genuinely new phenomenon, that constraint makes it difficult to publish. So the field cedes the territory, and it loses relevance as a result.

Prof. Sanders: You balance academic research with public writing and media engagement. How do you manage that tension?

Prof. Cappelli: I think of them as different audiences with different needs, and I do not find the tension as difficult as some people describe. On the academic side, the key is finding contexts where you can study something cleanly. My colleagues and I published a paper on what happens when someone receives a birthday card late – a study of the smallest possible workplace insult. The reason it was publishable was that we could find a context where we could isolate and measure the effect precisely. That is what academic audiences care about.

Learn more: Denise Rousseau on understanding the psychological contract

On the practical side, there are many important questions that simply are not being written about. I have published extensively in Harvard Business Review, which has around a million subscribers. If the topic matters and the evidence is there, people will find it.

On artificial intelligence specifically, I have stopped writing predictions. Predictions about AI and work have been wrong for years. What I am interested in now is what actually happens when organisations try to implement AI – the practical reality, not the forecast. Those studies are very hard to find, but they are the ones that would actually help managers make decisions.

The difficult part of having a public profile is that you occasionally get into arguments with people who have a strong personal stake in a particular answer. Our recent book, In Praise of the Office, examined the evidence on remote work and argued that 100 years of management research shows that in-person interaction matters. Hybrid arrangements can work, but they require deliberate design to create the connections that do not happen naturally when people are not together. That position made some people very unhappy. But I am not going to misrepresent what the research shows to avoid conflict.