The Uneven Geography of Busan’s Business Growth

Busan’s Suyeong district reported a 34.3 percent increase in an index of average business revenue. Excluding one neighborhood reduces growth across the remaining comparable areas to 12.4 percent. The discrepancy reveals how commercial expansion can be concentrated even as South Korea’s business population grows faster than its workforce.

Busan’s Suyeong district appeared to be enjoying a substantial commercial expansion. Between 2020 and 2024, an official measure of average annual revenue per business across comparable 100-meter grid squares rose 34.3 percent, outpacing the 23.2 percent increase reported for Busan as a whole. The district also recorded a 20.1 percent increase in the average number of registered businesses per square, almost twice the citywide rate. Those figures, published by the government’s Dongnam Regional Data Office on November 11, 2026, suggested a district adding businesses while its measured average revenue rose even faster.

A closer calculation alters that picture. Breeze in Busan separated the 63 comparable squares in Mangmi 2-dong from the 745 used in Suyeong’s revenue comparison. The increase across the remaining 682 squares was 12.4 percent, well below the district’s reported 34.3 percent. Mangmi 2-dong had registered an extraordinary 101.9 percent rise in its own average, enough to account for approximately 72.7 percent of the increase in Suyeong’s equally weighted grid-based index. That proportion describes the neighborhood’s contribution to a statistical average, not its share of the district’s actual additional sales.

The concentration is more remarkable within Mangmi 2-dong itself. Average revenue per business fell in 34 of its 63 comparable squares, even as gains elsewhere more than doubled the neighborhood’s mean. In Millak-dong, the opposite occurred: most comparable squares registered increases, but the neighborhood’s average declined. The geographic extent of improvement and the magnitude of the average were telling different stories. Suyeong’s published results captured genuine changes in registered commercial activity; their distribution showed why the headline rate could say so little about the prospects of businesses operating near one another.

This local pattern intersects with a broader transformation of the Korean business economy, although the two datasets measure different populations and periods. National statistics released in September showed that South Korea had 200,637 more small and midsize enterprises in 2024 than in 2023, while the number of people engaged in those enterprises increased by only 7,157. Almost the entire net expansion in business numbers was associated with an increase in one-person enterprises. The national figures expose a change in how commercial activity is organized; the Suyeong figures reveal how unevenly its recorded gains can be distributed across space. Together they put pressure on the assumption that an expanding business population or a rising local average necessarily indicates a similarly broad increase in economic opportunity.

The Neighborhoods Behind the Average

The Suyeong analysis divides the district into squares measuring 100 meters on each side. For each square, the statistical office calculated average annual revenue per registered business, then took an equally weighted average across the squares for which the two years could be compared. A square containing relatively few enterprises thus has the same weight as one containing many. The resulting measure is useful for detecting geographic variation, but it is neither the district’s aggregate revenue nor the sales history of a fixed group of firms. Its movement can reflect changes in the kinds of businesses present, their individual scale and the earnings recorded within each small area.

Of the district’s 1,065 squares, 745 had comparable revenue observations for 2020 and 2024. Across those squares, the reported average rose from 521.3 million won to 700.1 million won per business. Mangmi 2-dong’s 63 comparable squares started from a much higher level—1,508.2 million won—and reached 3,045 million won in 2024. Excluding them produces an average of approximately 430.1 million won for the remaining squares in 2020 and 483.5 million won in 2024. Those values, calculated from the agency’s published rounded figures, yield growth of 12.4 percent.

Mangmi 2-dong represented 8.5 percent of comparable squares but contributed almost three-quarters of the increase in the equally weighted district index. The calculation is especially sensitive to the neighborhood’s high starting revenue and subsequent rise. A conventional measure that summed the sales of every business would assign weight differently, and no district-wide total can be recovered from this grid-average series alone. The independent recalculation therefore identifies the source of the index’s concentration without claiming to have measured the distribution of actual sales proceeds among firms.

The direction of individual squares supplies a second, independent check on the interpretation of the mean. Across Suyeong’s 745 comparable squares, 412 recorded rising average revenue per business, 332 declines and one no change. Mangmi 2-dong’s extraordinary increase rested on 29 rising squares against 34 declining ones. Millak-dong reported a 40.1 percent increase in average business density, the strongest among Suyeong’s ten administrative neighborhoods, yet its revenue measure slipped 1.5 percent. Seventy-seven of Millak’s 128 comparable squares increased and 51 declined. Gwangan 2-dong, where average business density rose 29.6 percent, recorded a 7.3 percent reduction in its revenue measure; the comparable squares were almost evenly divided between gains and losses.

These patterns distinguish the spread of commercial activity from the size of its measured returns. A neighborhood can acquire additional businesses, broadening the number of places where commerce occurs, while its average sales per business declines. Equally, a small number of large changes can dominate a local average even where more locations are moving in the opposite direction. In Suyeong, five of the ten administrative neighborhoods recorded declines in their mean revenue measure despite the district’s strong reported growth. The disparity is not an abstract statistical curiosity. A policymaker relying on the district headline would be observing a different economic landscape from a proprietor weighing whether the customers and income available on a particular block could sustain another lease or employee.

The figures do not identify the businesses responsible for Mangmi 2-dong’s unusually large increase, and the series covers registered enterprises across industries rather than retailers alone. The 320 squares without comparable revenue observations are missing from the change calculation, not evidence of 320 closures. The starting year, 2020, was disrupted by the pandemic, while both years’ figures are expressed in nominal won. These conditions matter when interpreting the size of the growth rate. They do not erase the contrast among the locations actually measured: Suyeong’s average was pulled upward by an exceptional part of the district while many other squares and neighborhoods followed a different course.

A Growing Register, a Smaller Operating Unit

South Korea’s latest small-business figures reveal a comparable problem of scale, measured in enterprises rather than geographic squares. The Ministry of SMEs and Startups counted 8,499,552 small and midsize enterprises in 2024, up 2.4 percent from 8,298,915 in 2023. Their combined workforce, including working proprietors, rose from 19,117,649 to 19,124,806, an increase of approximately 0.04 percent. Enterprises with only one person accounted for 6,666,023 of the 2024 total, or 78.4 percent, after growing by 225,754 in a single year. The number of enterprises with two or more people fell by 25,117.

The distribution of labor between those categories explains the small national change more precisely than the headline totals. Every one-person enterprise contributes one person to the workforce count. The additional 225,754 people in that category were offset by a decline of 218,597 in the number recorded within enterprises with two or more people, leaving the net increase of 7,157. These are changes in the stocks of two size categories; a firm crossing from one category to another changes the totals, so the second figure is not a count of employees dismissed. What changed decisively was the proportion of the business population operating at the smallest measured scale. The average number of people engaged per SME fell from approximately 2.40 in 2021 to 2.25 in 2024.

The age profile complicates the idea that this expansion principally represents a wave of new entrepreneurial ventures. The number of firms operating for three years or less declined 3.7 percent in 2024, while businesses aged four to seven years increased 4 percent and those older than seven years grew 6.2 percent. Age-group stocks cannot reveal the number of openings or closures, because businesses pass from one bracket to another. They nevertheless show that the youngest part of the enterprise population was contracting at a time when the register as a whole was expanding.

The revenue arithmetic points in the same direction. Combined SME turnover reached approximately 3,324 trillion won in 2024, up just 0.7 percent. With enterprises growing 2.4 percent, annual sales averaged about 391 million won per registered firm, down from approximately 398 million won the previous year. Dividing the same turnover by the number of people engaged produces a small increase of around 0.7 percent in nominal sales per person. The difference reflects the denominator: business units multiplied much faster than the labor recorded within them. Neither average measures owners’ income or the real value workers added, and a shift in the mix of firms can change both without an equivalent deterioration at businesses operating in consecutive years.

The economic possibilities behind these figures are materially different. A specialist enterprise may serve a larger market through digital systems, subcontracting and efficient production without expanding its permanent payroll. Another operator may remain alone because the turnover available after premises and purchased inputs cannot finance an employee. Size is therefore an incomplete proxy for productive capacity. The national data capture a reorganization toward more one-person businesses; the distinction between commercial efficiency and constrained self-employment depends on earnings, work arrangements and market access that the registration count does not contain.

Commerce Beyond the Storefront

Retail provides a useful test of any assumption that smaller business units necessarily mean fewer people working in an industry. In the national statistics agency’s provisional 2024 Service Industry Survey, retail establishments gained approximately 30,000 people engaged, even as wholesale and intermediary trade lost about 52,000. The combined wholesale and retail category, including motor-vehicle sales, recorded a decline of roughly 23,000. The survey counts establishments rather than the enterprises in the ministry’s SME register, so its figures describe a separate population. Within that population, the employment movements diverged sharply between the businesses selling directly to consumers and those moving goods through wholesale channels.

Retail establishments increased 4.4 percent, from approximately 1.06 million to 1.11 million, while their combined workforce grew 1.5 percent to about 2.02 million. That reduced the average workforce per establishment from 1.88 to 1.83 people. Retail sales rose 1.3 percent to approximately 524 trillion won, leaving average nominal sales per establishment about 3 percent lower, at 474 million won. The number of operating units was growing faster than both the recorded work and the turnover they collectively supported. An expanding retail industry was becoming more fragmented in its establishment-level averages without undergoing an overall reduction in people engaged.

Distribution channels were evolving alongside that fragmentation. The share of retail establishments using delivery or parcel-based sales increased from 48.2 percent in 2023 to 51.9 percent in 2024; the share conducting transactions through digital platforms rose from 26.7 to 28.1 percent. These are the proportions of establishments using each channel, rather than percentages of their sales. Still, the expansion changes where work can be organized. Payment services, inventory systems and delivery networks allow some sellers to operate beyond the demand immediately surrounding their premises. A purchase can be made locally, received from a seller elsewhere and fulfilled by workers employed in logistics rather than in the retail establishment.

The Bank of Korea’s July 2025 research documented how unevenly that change has unfolded. Growth in nonstore retailing was concentrated in the Seoul metropolitan area, while traditional store-based retailers outside the capital region faced greater pressure. Larger businesses and stores that had adopted online sales generally performed better than competitors with narrower distribution capabilities. The research also found substantial employment growth in parcel delivery. Those observations do not assign responsibility for the separate 2024 survey’s employment changes to any single technology. They establish that the reach of a business, the place a consumer spends money and the location of the work serving that purchase increasingly differ.

The effects on a district such as Suyeong depend partly on whether its enterprises can draw customers and income into the local economy or instead compete for spending already circulating there. A restaurant or service business drawing visitors from beyond the neighborhood can add demand; a new retailer serving the same customers as nearby merchants may redistribute it. Neither relationship can be read from the number of businesses per square. The economic geography of commerce extends beyond the physical geography of the establishments, making the distribution of income and the capacity to sustain work more consequential than a simple count of storefronts.

The Income Behind the Sales

The accounts of Korean restaurants make the cost of sustaining that work tangible. A government survey covering 3,138 food-service establishments reported average annual sales of 255.26 million won for 2024. Operating profit averaged 22.32 million won, or 8.7 percent of turnover, after food purchases, labor, premises and other operating expenses. Restaurants are one part of the broader commercial economy rather than a representative sample of all the enterprises in Suyeong’s grid analysis. Their cost structure nonetheless shows why higher recorded sales cannot be translated directly into a proprietor’s income or capacity to employ other people.

Food ingredients absorbed 40.7 percent of sales and wages paid to employees another 14.3 percent. The survey separately assigned 13.3 percent to the proprietor’s own labor and 3.3 percent to the work of family members. Together, those owner and family labor costs represented 16.6 percent of turnover, exceeding the share paid to hired employees. They are accounting estimates of the economic cost of labor, rather than necessarily wages taken home in cash, and were included in expenses before the reported operating profit was calculated. A proprietor’s contribution to a restaurant therefore supports both its operation and an economic claim on its earnings, even where the cash payroll is limited.

The relationship between rent and hired labor varied substantially with turnover. Restaurants reporting less than 50 million won in annual sales devoted 15.3 percent of their revenue to rent and 6.9 percent to employee wages. Among businesses generating at least 500 million won, rent represented 6.1 percent and employee wages 17.1 percent. These groups contain different kinds of restaurants and are not successive stages in a tracked firm’s growth. Yet the disparity shows how premises can absorb a markedly larger share of modest turnover while comparatively little is allocated to paid staff. Bringing in another employee requires demand and reliable earnings sufficient to carry a recurring wage obligation, not simply the continued existence of the business.

Competition also bears directly on that calculation. In the ministry’s separate survey of small enterprises for 2024, 61 percent of respondents cited intensifying competition among their business difficulties and 33.5 percent cited declining commercial districts. These are reported business concerns, not measurements of actual sales losses. Read alongside the restaurant accounts, they give substance to an operating constraint: a business may be able to pay for space, supplies and the proprietor’s own work while remaining too exposed to fluctuations in customers to add another regular wage. More efficient independent operators can face the opposite situation, expanding sales through technology and outside services without any commercial need for a larger payroll.

Measuring the Economy That Remains

In September 2026, the Ministry of SMEs and Startups proposed a change to the information used to identify beneficiaries of small-business support. The Cabinet-approved legislative amendment would permit a closer combination of tax, income, expense and labor records, reflecting the fact that one proprietor may operate several unmanned shops while a salaried worker may maintain a separate online business. The proposal was not an implemented system. Its premise, however, exposes a consequential weakness in treating each business registration as an equivalent unit of need, livelihood or productive expansion.

The same distinction applies to local commercial policy. Incentives that help an enterprise open can fill vacant premises and reduce an entrant’s costs, but their effect on the surrounding market depends on the additional spending the enterprise generates and the revenue it draws from existing competitors. Assistance aimed at a productive independent firm with access to new customers addresses different circumstances from income protection for a proprietor operating with persistently low earnings. Support intended to increase employment has another requirement again: businesses must have sufficiently durable demand and margins to sustain the wages and hours being encouraged. A single business-count target cannot measure the success of all three.

Suyeong’s unusually strong headline growth rate makes this distinction particularly visible. An average dominated by Mangmi 2-dong describes a measurable concentration of commercial performance; within that neighborhood, a majority of comparable squares nonetheless recorded declining average revenue per business. Elsewhere in the district, new businesses and rising sales occurred in combinations that sometimes moved in opposite directions. A better account of local development would follow continuing firms as well as new entrants, distinguish revenue from income after labor and premises costs, and measure paid employment through wages and hours. Only then could the increase in registered activity be separated into expanding local demand, more efficient production and the redistribution of sales among operators.

Suyeong’s revenue index assigns each comparable square equal weight; the national register counts each enterprise once. Households and employees experience commerce through a different accounting: the earnings remaining after rent, inputs and labor, and the work those earnings can reliably support. Suyeong’s sharp contrasts make that accounting urgent. A district whose growth measure depends so heavily on one neighborhood needs to know where new customers and wages are appearing, and where enterprises are dividing spending already available to the businesses around them.