Technical efficiency measures how well a producer converts inputs into outputs. A firm, hospital, or factory is technically efficient when it squeezes the maximum possible output from its resources, or equivalently, uses the fewest possible resources to produce a given output. A score of 1.0 (or 100%) means the operation sits right on the frontier of what’s physically possible with current technology. Anything below that number represents waste.
The Core Idea: Production Frontiers
Imagine plotting every factory in an industry on a graph, with inputs on one axis and outputs on the other. The best performers form a boundary line called the production frontier. This frontier represents the maximum output achievable for any given level of input, using the best available technology. Factories sitting on that line are technically efficient. Factories falling below it are not, and the distance between where they sit and the frontier tells you exactly how much waste exists in their process.
The concept traces back to the economist Tjalling Koopmans, who defined it in 1951 with an elegantly simple rule: a producer is technically efficient when it’s impossible to reduce any input or increase any output without simultaneously increasing another input or decreasing another output. In other words, there’s no slack anywhere in the system. Every resource is pulling its weight.
Michael Farrell built on this in 1957 by introducing a way to actually measure how far a producer falls from the frontier. His framework turned the concept from a theoretical idea into something you could calculate with real data, and it remains the foundation of efficiency measurement today.
How It Differs From Allocative Efficiency
Technical efficiency and allocative efficiency are related but answer different questions. Technical efficiency asks: “Are you getting the most out of your resources?” Allocative efficiency asks: “Are you using the right mix of resources given their prices?” A hospital might be technically efficient, running at full capacity with no idle staff or empty beds, but allocatively inefficient if it employs too many specialists relative to general practitioners for the patient mix it serves. Both matter for overall economic efficiency, but technical efficiency focuses purely on the physical relationship between inputs and outputs, ignoring costs and prices entirely.
Input vs. Output Orientation
There are two ways to think about improving technical efficiency, and the distinction matters depending on the situation. Input-oriented efficiency asks: “By how much can we shrink our inputs while still producing the same output?” Output-oriented efficiency asks: “By how much can we grow our output using the same inputs?”
If you run a public school system where enrollment is fixed, input orientation makes more sense because you’re looking to deliver the same education with fewer resources. If you manage a factory that can sell everything it makes, output orientation is the better lens because you want more product from the same machinery and labor. Interestingly, comparing the two measures for the same operation reveals something useful: when input-oriented efficiency is higher than output-oriented efficiency, the operation is experiencing increasing returns to scale and would benefit from getting bigger. When the reverse is true, it should consider scaling down.
How Technical Efficiency Gets Measured
Two major methods dominate efficiency measurement, and they approach the problem from completely different directions.
The first, Data Envelopment Analysis (DEA), is a mathematical programming technique that doesn’t assume any particular shape for the production frontier. It builds the frontier directly from the data by connecting the best-performing units, then measures how far everyone else falls from that boundary. DEA works well when you’re comparing similar operations, like branches of a bank or hospitals in a network, because it handles multiple inputs and outputs without requiring prices. The trade-off is that it attributes all deviations from the frontier to inefficiency, with no allowance for measurement error or random bad luck.
The second, Stochastic Frontier Analysis (SFA), takes a statistical approach. It estimates the production frontier using regression and splits the gap between a producer and the frontier into two pieces: one representing genuine inefficiency, and one representing random noise like data errors, weather disruptions, or equipment failures beyond anyone’s control. This separation is SFA’s major advantage, but it requires the analyst to assume a specific mathematical form for the production function, which may not always fit reality.
Studies frequently find that DEA and SFA produce different efficiency scores even when applied to the exact same data. The discrepancies come from two sources: the methods shape the frontier differently, and they measure the distance to that frontier differently. Neither is universally “better.” The right choice depends on the data available, the number of operations being compared, and whether random variation is a serious concern.
What the Scores Look Like in Practice
Technical efficiency scores range from 0 to 1, where 1 means fully efficient. To make this concrete, consider U.S. Department of Defense hospitals studied from 2010 to 2021. Their average technical efficiency score over that period was 0.950, meaning the typical hospital could have produced the same care using roughly 5% fewer resources. By 2021, the average had climbed to 0.991, nearly closing the gap to the frontier entirely. Scale efficiency, which captures whether hospitals were operating at the right size, reached a perfect 1.000 in both 2020 and 2021.
These numbers illustrate a common pattern: most organizations in a sector cluster somewhere in the 0.80 to 0.99 range, with the gap representing real but often modest room for improvement. A score of 0.85 means an operation is wasting about 15% of its inputs relative to the best performers doing comparable work.
Technical Efficiency on the Factory Floor
In manufacturing, the most widely used proxy for technical efficiency is Overall Equipment Effectiveness (OEE). OEE multiplies three factors together: availability (how much scheduled time the equipment actually runs), performance (how close it runs to its maximum speed), and quality (what percentage of output is defect-free). A perfect OEE score of 100% means a machine produces only good parts, as fast as possible, with zero downtime.
OEE breaks the abstract concept of technical efficiency into pieces a plant manager can act on. If availability is dragging the score down, the problem is unplanned stops or excessive changeover time. If performance is the weak link, the machine is running below its rated speed or experiencing frequent micro-stops. If quality is low, defective products are consuming resources without producing usable output. Each of these maps directly onto the economist’s definition: inputs being consumed without generating corresponding outputs.
Why It Matters Beyond Economics Textbooks
Technical efficiency has real consequences for resource allocation decisions. Governments use efficiency scores to identify underperforming public hospitals, schools, and utilities, then target interventions where the gap between actual and potential performance is widest. In agriculture, efficiency analysis reveals which farms produce less than they should given their land, labor, and equipment, pointing toward training programs or technology adoption rather than simply adding more resources.
For businesses, the concept reframes a common instinct. When output is too low, the natural response is to buy more equipment or hire more staff. But if the operation is technically inefficient, additional inputs will be partially wasted too. The better first step is closing the gap to the frontier with existing resources, then scaling up once the operation runs cleanly. A firm operating at 0.80 efficiency that doubles its inputs won’t double its output. It’ll get roughly 1.6 times the output, with the same proportional waste baked in. Fixing the inefficiency first means every future dollar of investment works harder.