IT Capability and Organizational Performance: What the Research Shows

In the 1980s and early 1990s, economists noticed something strange. Organizations were spending enormous amounts on computers, and the productivity statistics were not moving. This became known as the "productivity paradox," and it produced a serious research debate in both economics and information systems. The puzzle was simple: if IT is valuable, why does economy-level data not show it?

Brynjolfsson (1993) identified four possible explanations. Mismeasurement, because traditional productivity statistics were not designed to capture service-sector or information-based output well. Time lags, because the productivity benefit from a technology may take years to materialize as organizations learn to use it. Redistribution, because gains to individual firms can cancel out at the aggregate level if competitive markets pass value to consumers rather than to profit. And mismanagement, meaning organizations were not always using IT well. The paradox was real, but its interpretation required care.

At the firm level, Brynjolfsson and Hitt (1996) found that IT investment did correlate with higher output when you looked at individual companies rather than economy-wide aggregates. IT created value, but that value often appeared as consumer surplus rather than firm profit in competitive markets. This distinction matters: generating value is not the same as capturing it.

What I find most interesting about this debate is where it eventually landed. The answer was not "IT spending does not matter" and it was not "IT spending always pays off." The answer was that it depends on what the organization does with the technology. That is a much less satisfying answer for a CFO trying to justify a budget line, but it is the defensible one.

This is where Bharadwaj (2000) became important for the IS field. She found that firms with high IT capability had significantly higher profit ratios and lower cost ratios than a matched sample of other firms. But her definition of IT capability was not simply "how much did they spend on IT." IT capability in her study had three components: IT infrastructure (the technical platform, the hardware, the networks), IT human resources (the people who design, manage, and operate IT), and IT-enabled intangibles (things like organizational knowledge, synergy, and embedded routines that were partly the product of IT use over time). The study is cited in the IS research literature as an important empirical demonstration that the unit of analysis matters enormously. Capability, not spending, is what predicts performance.

This connects directly to why Carr's (2003) "IT Doesn't Matter" argument is only partly right. Carr's claim was that IT, as it standardizes and becomes commoditized infrastructure, can no longer be a source of competitive advantage. He was right that commodity infrastructure fails the VRIN test, as I wrote about in my post on resource-based view and IT competitive advantage. If your competitors can buy the same software from the same vendor and implement it in roughly the same way, the purchase is a table stake, not a differentiator.

But Carr's argument applied most cleanly to IT resources and not to IT capability in Bharadwaj's sense. Two organizations can spend the same dollar amount on the same enterprise software and end up with radically different capabilities, because capability is not just the technology. It is the people, the processes, the accumulated routines, and the organizational knowledge about how to use the technology effectively inside a specific context. That combination can be hard to observe and hard to replicate. According to Gartner research on technology business management, the difference between organizations that realize value from their technology investments and those that do not is rarely about the technology itself (see https://www.gartner.com/en/newsroom for Gartner's current commentary on this topic). I am hedging Gartner's specific figures here, but the pattern is consistent with what the IS academic literature shows.

The measurement problem has never been fully resolved, and I think it is worth sitting with. Operationalizing "IT capability" in a survey or a financial dataset is genuinely hard. Most studies end up using IT investment as a proxy, which reintroduces the problem that Bharadwaj's work was trying to move past. If you measure IT capability by asking how much the firm spends on IT, you are back to the productivity paradox territory. If you try to measure capability more directly, you get into the complexity of assessing human IT skill, organizational routines, and intangible knowledge assets, all of which are real but difficult to quantify.

Melville, Kraemer, and Gurbaxani (2004) tried to address the measurement problem at a different level. Their Business Value of IT (BVIT) model specified the chain more precisely: IT resources plus complementary organizational resources produce improved business processes, which then produce organizational performance, which then feeds competitive outcomes. The chain is the mechanism. Skipping it and jumping directly from IT investment to firm performance is where most oversimplified claims about IT value go wrong.

My read of this literature is that the productivity paradox was mostly a measurement and aggregation artifact, resolved empirically at the firm level by the mid-1990s. But the substantive question it raised, under what conditions does IT actually contribute to organizational performance, is still not fully answered. Bharadwaj's three-part definition of IT capability is probably the most useful framing I have encountered for thinking about why two organizations with similar technology spending end up with very different outcomes.

The practical implication is uncomfortable for organizations that prefer simple answers: buying the right software is not sufficient. Building the capability to use it well, over time, with the right people and the right organizational conditions, is what the evidence actually supports.