Shift-Share Analysis: Formula, Example, and How to Run One

I once spent $127,000 opening three new sales territories. And I was SO proud of the numbers.

Then a colleague ruined my week with one question.

She asked, “How much of that growth is you, and how much is just the economy?” I didn’t know. I had celebrated 45% growth in one region as pure team execution. When I finally ran the math, only about 7 points of it came from us. The rest was national momentum and a lucky industry mix.

That stung. But it taught me the one method I now run before any territory decision: shift-share analysis.

So let me save you the expensive version of this lesson. Here’s what shift-share analysis is, the three components it splits growth into, how to run one step by step, and a worked example with clean numbers. Let’s get into it.

📌 TL;DR: Shift-share analysis splits a region's observed change into three parts: the National Growth Effect (the whole economy rising), the Industry Mix Effect (your region happens to sit in fast- or slow-growing industries), and the Regional Competitive Effect (what the region itself actually did). Observed Change = National Growth Effect + Industry Mix Effect + Regional Competitive Effect. The first two are tide and address. Only the last one is you.

What Is Shift-Share Analysis?

Shift-share analysis is a regional economics method that decomposes change in a variable, usually employment, into three components: national growth, industry mix, and regional competitiveness. It answers one question. How much of this change did the region actually earn?

Think about a metro area growing employment 9% over five years. Nice headline. But three very different forces could be hiding inside it.

Maybe the whole national economy grew, and the region rose with the tide. Maybe the region is packed with industries that boomed everywhere. Or maybe something local (talent, infrastructure, policy, hustle) genuinely beat the odds. Shift-share analysis separates those three so you stop crediting luck and stop blaming teams stuck in shrinking industries.

Because of that, it’s a standard first tool in regional economic analysis and economic development planning. Analysts use it on jobs, GDP, or income. Same logic every time.

The Three Components of Shift-Share Analysis

Shift-share analysis breaks observed change into three effects that add up exactly to the total. That additive property is the whole charm. Nothing gets lost, nothing gets double counted.

Component	Classic name	What it measures	Is it “you”?
National Growth Effect	National share	What the region would gain if it simply grew at the overall national rate	No. It’s the tide.
Industry Mix Effect	Proportional shift (industrial mix)	The bonus or penalty from being concentrated in fast- or slow-growing industries	No. It’s your address.
Regional Competitive Effect	Differential shift	The growth left over after national and industry forces are stripped out	Yes. This is local performance.

The formula in words → Observed Change = National Growth Effect + Industry Mix Effect + Regional Competitive Effect. Now each piece, plainly.

National Growth Effect

The National Growth Effect is what a region would gain if it grew at the exact national growth rate, nothing more. It’s the baseline every region shares.

You calculate it by multiplying the region’s starting employment by the overall national growth rate. That’s it. In an expansion year, this national growth component can quietly explain most of what looks like regional success. So a big chunk of any boom-year brag was never local.

Industry Mix Effect

The Industry Mix Effect measures whether the region’s industry composition helped or hurt. Heavy in software during a software boom? Tailwind. Heavy in a declining sector? Headwind you didn’t choose.

For each industry, you take the region’s starting employment in that industry and multiply by the gap between that industry’s national growth rate and the overall national rate. Then you sum across industries. A positive industrial mix effect means the region is concentrated in national winners. And a negative industrial mix means structural drag, not local failure.

Regional Competitive Effect

The Regional Competitive Effect is the residual, meaning the observed change minus the national and industrial mix effects. And this is the only piece that reflects local performance.

Industry by industry, it compares the region’s growth rate to that industry’s national growth rate. Positive means local firms beat their national peers. Negative means they lagged, even if the headline still looks fine. This differential shift is the number worth managing to. Everything else is context.

How Do You Run a Shift-Share Analysis? (Step by Step)

You run one by picking a period, pulling industry employment for the nation and the region, computing the three effects, and reading the residual. Here’s the sequence I follow.

• Pick your variable and period. Employment is standard. Choose a clean base year; a distorted one (like 2020) warps every effect downstream.
• Pull the data on consistent industry codes. In the US that means NAICS codes (the standard industry classification). The BLS Quarterly Census of Employment and Wages and Census County Business Patterns cover employment by industry and county. For GDP and income, the BEA regional accounts go deeper. In Europe, Eurostat’s regional statistics play the same role.
• Compute the National Growth Effect. Regional base employment × the national growth rate.
• Compute the Industry Mix Effect. Per industry: base employment × (industry national rate – overall national rate). Sum it.
• Compute the Regional Competitive Effect. Per industry: base employment × (regional industry rate – national industry rate). It also equals observed change minus the first two effects.
• Sanity-check the sum. The three effects must add to the observed change exactly. If they don’t, an input is wrong.

One honest warning from experience: your answer is only as trustworthy as your inputs. Small counties suppress data for privacy, NAICS codes shift between vintages, and one mismatched region quietly corrupts everything. So real data preparation (standardizing codes, handling suppressed cells, aligning geographies) isn’t the boring part. It’s the part that decides whether the analysis is true.

At the regional level, data quality is where most analyses quietly break. Some counties hide employment counts when only a few firms operate in an industry. So I aggregate up to a metro area or a broader industry code. Enough to keep the math honest without exposing any single company.

A Worked Example (Hypothetical Numbers)

That’s the idea. Now the math, with a fully hypothetical region and round numbers so nothing distracts.

Say our made-up region has two industries. Manufacturing starts with 6,000 jobs and software starts with 4,000, so 10,000 total. Five years later the region holds 10,900 jobs. That’s +900, a 9% headline. Meanwhile, suppose the national economy grew 5% overall, national manufacturing grew 2%, and national software grew 12%.

Industry	Base jobs	Actual change	National effect	Industry mix	Competitive effect
Manufacturing	6,000	+100	+300	-180	-20
Software	4,000	+800	+200	+280	+320
Total	10,000	+900	+500	+100	+300

Walk one row. Manufacturing’s national effect → 6,000 × 5% = 300 jobs. Its industrial mix → 6,000 × (2% – 5%) = -180 jobs, a structural headwind. And its actual gain was only 100 jobs, so the competitive effect is 100 – 300 – (-180) = -20. Local manufacturers slightly lagged their national peers.

And the check works: 500 + 100 + 300 = 900. Exactly the observed change.

So the 9% headline splits into a 5-point tide, a 1-point lucky address, and 3 points of real regional strength, almost all of it in software. “We grew 9%” and “our software firms are outcompeting the nation while manufacturing slips” are very different stories. Shift-share analysis is how you get the second one.

Strengths and Limitations of Shift-Share Analysis

Headline growth lies. The split doesn’t, but it has limits worth knowing before you present it to anyone.

Strengths first. It’s simple, cheap, and runs on free public data. The three effects add up exactly, which makes it easy to explain to non-economists. And as a first diagnostic, it’s hard to beat: one afternoon of work tells you whether a region’s story is tide, address, or execution.

Now the honest limits. Shift-share analysis is descriptive, not causal. It tells you a competitive effect exists, not why. Results also depend on your base year and how finely you slice industries; change either and the numbers move. The classic static form compares just two points in time, which is why analysts built dynamic shift-share, a variant that recomputes the effects year by year. And the residual soaks up measurement error along with genuine local performance. The Wikipedia entry on shift-share analysis covers the classic and dynamic variants well.

My sanity-check routine: run it with two different base years and two aggregation levels. If the story survives all four runs, I trust it. If it flips, I trust none of them. Worth knowing too: econometricians borrowed this exact logic to build Bartik instruments for causal research, and the AEA’s practical guide to shift-share instruments is the reference if you’re headed that way.

How I Use Shift-Share in B2B Planning

So why does a marketing manager care about a regional economics method? Because territories are little regional economies.

For quotas, I set targets off expected change (national effect plus industry mix) instead of last year’s raw number. A team in a structurally lucky region gets a higher bar. A team fighting a bad industrial mix doesn’t get punished for gravity. For evaluating managers, I grade on the competitive effect alone, because that’s the part they control.

The prep work is the same as the economist’s, just messier. You’re mapping accounts to the right industry and location, which leans on clean company data and reliable company identifiers. Sloppy data matching puts a firm’s employment in the wrong region, and the whole decomposition tilts.

💡 Field-tested move: Before you praise or panic about any region, run the split. A "weak" 4% region with a positive competitive effect is often a better bet than a "strong" 15% region riding pure national luck. Reward the residual, not the headline.

Common Mistakes (I Made Most of These)

I’ve botched shift-share analysis in every way available. Here are the ones that cost me real money and credibility.

First, I trusted a single base year. My 2020 baseline was pandemic-warped, so every effect downstream was noise. Now I average across several years. Second, I mixed NAICS vintages without a crosswalk, which quietly double-counted an industry that got reclassified. Third, and worst, I matched companies to regions on fuzzy names alone and put a large employer in the wrong county. One bad match, one wrong conclusion, one embarrassing meeting.

Two more I see constantly: reading the headline instead of the residual, and treating a positive competitive effect as causal proof of good strategy. It’s evidence. It’s not a verdict.

The pattern behind all of them? Garbage in, confident garbage out. Slow down on the inputs and most of the danger disappears.

🧠 Keep in mind: Shift-share analysis describes what happened; it never proves why. Use it to find the regions worth investigating, then go find the actual cause on the ground before you spend money on the answer.

Frequently Asked Questions

What is an example of a shift-share analysis?

A hypothetical region grows employment 9%; shift-share splits that into 5 points of national growth, 1 point of industry mix, and 3 points of regional competitiveness. So most of the headline came from the tide and the region’s industry composition, not a unique local edge. Real analyses run the same decomposition on published employment data for an actual region and period.

How do you interpret shift-share analysis?

Read the three effects separately: national means the economy helped, industry mix means the region’s industries helped, and the competitive effect means the region itself outperformed. The Regional Competitive Effect is the one that reflects local execution. A big national effect with a tiny competitive effect means the region is riding a wave. And waves go out.

What is the difference between shift-share analysis and location quotient?

A location quotient measures how concentrated a region is in an industry at one point in time; shift-share analysis measures how the region changed over time and why. They pair well. Highly concentrated industries usually drive the biggest industrial mix effects, so I compute quotients first to see which industries will dominate the decomposition.

What data do I need for shift-share analysis?

You need employment by industry for both the nation and your region, on consistent industry codes, at two points in time. BLS QCEW, Census County Business Patterns, and BEA regional accounts cover the US for free, and Eurostat covers European regions. The hard part is keeping codes and geographies consistent across the two dates.

What is a Bartik shift-share instrument?

A Bartik instrument predicts regional growth by combining a region’s starting industry shares with national industry growth rates. Econometricians use it as an instrumental variable, a stand-in that isolates outside-driven demand shifts, for causal questions descriptive shift-share can’t answer. Same ingredients as classic shift-share, much stricter statistical job.

Why did my competitive effect come out negative?

A negative competitive effect means the region grew less than its national and industry conditions predicted, so it’s underperforming its context. Sometimes that’s a real problem, like weak execution or a thinning talent pool. Sometimes it’s temporary disruption or a data artifact. So check whether the negative sign persists across several base years before acting on it.

Can I use shift-share for something other than regions?

Yes. The same logic works for sales territories, customer segments, product lines, and even population projections. You swap “national economy” for your total book of business and “industry” for segment, then read the residual as segment-specific performance. Demographers use a shift-share method for population projection the same way. The math is identical; only the labels change.

It’s Time to Split the Tide From the Team

Next time a regional number looks amazing (or awful), don’t react to the headline. Run the split. Twenty minutes with free public data will tell you how much was tide, how much was address, and how much was real.

Because once you’ve seen a 45% “win” shrink to 7 points of actual performance, you never read a growth chart the same way again. Tell me in the comments: what region are you going to run this on first?

---