Thailand's Most Volatile Decade for Company Formation, 2026
In short
The 2010s were Thailand's most volatile decade for new company formation, with a coefficient of variation of 32.4 percent, more than three times the 2020s figure of 9.7 percent. Registrations in the 2010s climbed unevenly from 41,404 in 2010 to 111,284 in 2019. The 2020s, despite the pandemic and a sharp reversal after 2021, held a tighter relative range around a much higher base near 126,000 a year. The earliest years on record, 2007 to 2009, look calmer still at 6.4 percent, but that reading rests on only three years of data.
Which decade of Thai company formation was the most volatile?
Measured by coefficient of variation, the 2010s take the title. Annual new company registrations from 2010 through 2019 carried a coefficient of variation of 32.4 percent, against 9.7 percent for the 2020s so far and 6.4 percent for the fragment of the 2000s the registry covers, 2007 through 2009 (DBD, 2026). Coefficient of variation divides the standard deviation of a series by its mean, so it measures relative swing rather than raw headcount. Registrations in the 2010s rose from 41,404 in 2010 to 111,284 in 2019, but not smoothly. Growth jumped 21.8 percent in 2013 and 22.6 percent in 2017, then slowed to single digits in the years between. The 2020s tell a different story. Formation ran from 122,427 in 2020 to a record 151,205 in 2021, then fell 20.7 percent to 119,923 in 2022. Those raw swings look dramatic, but they happened against a decade average more than 55,000 companies higher than the 2010s average, so the relative dispersion is much smaller. By coefficient of variation, the 2010s were about 3.3 times as volatile as the 2020s.
Coefficient of variation in new company registrations, by decade
- 2000s, 2007 to 20096
- 2010s, 2010 to 201932
- 2020s so far, 2020 to 202510
Figures in percent coefficient of variation.
The 2010s stand well clear of both neighboring windows. The 2020s reading covers 2020 through 2025 only, since 2026 is not yet complete.
Decade by decade formation volatility
| Decade | Years in sample | Mean annual formations | Standard deviation | Coefficient of variation |
|---|---|---|---|---|
| 2000s, 2007 to 2009 | 3 | 32,897 | 2,105 | 6.4% |
| 2010s, 2010 to 2019 | 10 | 70,689 | 22,931 | 32.4% |
| 2020s so far, 2020 to 2025 | 6 | 126,048 | 12,206 | 9.7% |
Annual new company registrations grouped by decade, with mean, standard deviation and coefficient of variation, as of 17 Jul 2026.
Why decade level volatility matters
The instinct is to call the pandemic years the roughest stretch in Thailand's company register. Thailand's GDP contracted sharply through 2020 and into early 2021, a shock deep enough to make choppy formation look like the obvious outcome (World Bank, 2021). The instinct does not survive contact with the coefficient of variation. A trade credit provider or a lender who treats a single year move in 2020s formation as a break in trend is reading more noise into the data than the decade actually contains, since 2020 to 2025 has held closer to its own average than the 2010s ever did. The 2010s pattern of repeated double digit swings around a rising trend is the harder one to forecast against, and it is the one that should set the benchmark for what a genuinely volatile stretch looks like. The 2007 to 2009 window is worth a caveat rather than a conclusion. Three years produce a low coefficient of variation almost by construction, not because formation was calm before the registry's data begins in 2007.
Definitions
- Coefficient of variation
- The standard deviation of a series divided by its mean, expressed as a percent. It lets you compare relative volatility across periods that sit at very different scales.
- Standard deviation
- A measure of how far the values in a series typically sit from the average of that series, in the same units as the underlying data.
- Founding year
- The registry field recording the year a company was formally registered with the Department of Business Development. This analysis groups companies by that year.
- Population standard deviation
- A standard deviation computed by treating every year in a decade as the full set of interest, dividing squared deviations by the count of years rather than one less than that count.
Assumptions
- The 2020s window used for the headline comparison covers 2020 through 2025. The 2026 row is excluded from every decade calculation because the year is still in progress and its count is not comparable to a completed year.
- The 2000s window is limited to 2007 through 2009, the earliest years the market_new_companies_insight table carries. It is not a full decade, and its coefficient of variation should be read as a data limitation rather than a settled result.
- Coefficient of variation is sensitive to trend as well as to noise. A decade with strong sustained growth, such as the 2010s, will show a higher reading than a flatter decade with similar percentage swings, so the measure captures both direction and dispersion together.
- Standard deviation here is computed as a population statistic across each decade's own years, not as a sample estimate of some larger population.
Methodology
Annual new company counts are grouped by founding year from the DBD registry via the market_new_companies_insight aggregate table, with rows computed at 2026-07-17. Coefficient of variation is the population standard deviation of each decade's annual counts divided by the decade's mean, expressed as a percent. This measures relative dispersion, so decades with different bases can be compared directly. The 2000s reflect only 2007 through 2009, the earliest years the table carries. The 2020s cover 2020 through 2025. The year 2026 is excluded from every decade calculation because registrations for it are still being recorded and the count would understate a full year.
How we verified this
- 2010s mean recomputed as the sum of the ten annual counts from 41,404 to 111,284, equal to 706,885, divided by 10 years, equal to 70,688.5.
- 2010s population standard deviation recomputed from those ten counts as 22,931.3, giving a coefficient of variation of 22,931.3 divided by 70,688.5, equal to 32.4 percent.
- 2020s mean for 2020 through 2025 recomputed as the sum of the six annual counts, 756,290, divided by 6 years, equal to 126,048.3, with a population standard deviation of 12,206.5, giving a coefficient of variation of 9.7 percent.
- 2007 to 2009 mean recomputed as 98,691 divided by 3 years, equal to 32,897, with a population standard deviation of 2,104.6, giving a coefficient of variation of 6.4 percent.
- 2010s to 2020s ratio confirmed as 32.4 divided by 9.7, equal to 3.3, so the 2010s were roughly three and a third times as volatile as the 2020s by this measure.
- The year over year moves cited in the text were checked against the raw counts: 2013 growth of 10,804 companies over 49,466 equals 21.8 percent, 2017 growth of 16,556 over 73,163 equals 22.6 percent, and the 2021 to 2022 drop of 31,282 over 151,205 equals negative 20.7 percent.
References
- Department of Business Development. (2026). Registered juristic persons and new company formation aggregates. Ministry of Commerce, Thailand. https://www.dbd.go.th/
- World Bank. (2021). Thailand economic monitor, July 2021: The road to recovery. World Bank Group. https://www.worldbank.org/en/country/thailand/publication/thailand-economic-monitor-july-2021-the-road-to-recovery
Frequently asked questions
Which decade had the most volatile Thai company formation?
The 2010s, with a coefficient of variation of 32.4 percent in annual new registrations, more than three times the 2020s reading of 9.7 percent.
Why use coefficient of variation instead of raw year to year swings?
Coefficient of variation divides the standard deviation of a series by its mean, so it compares relative dispersion across periods with very different scales. The 2020s swung by tens of thousands of companies in absolute terms, but against a much higher base those swings represent a smaller relative move than the swings of the 2010s.
Was the 2000s decade really the calmest?
The registry table only carries 2007 through 2009 for that decade, three years of data, so its 6.4 percent coefficient of variation is not a fair match against the full ten year 2010s reading or the six year 2020s reading. Treat it as a data limitation, not a finding about the whole decade.
Does the 2020s figure include 2026?
No. The 2026 row is excluded from every decade calculation because the year is still being recorded and its count would drag the average down without reflecting a full year of formation.
Cite this page
Chatpong L. (2026). Thailand's Most Volatile Decade for Company Formation, 2026. Max Data Insights. https://maxdatathailand.com/insights/thailands-most-volatile-decade-for-company-formation-2026
Data & corrections
Download the data (CSV)Spot something off, or want to ask about the data? Email insights@maxsolutions.co.th. We publish corrections.
Written by
Chatpong L.Founder, Max Data, Max Solutions Co., Ltd.
Chatpong L. leads Max Data, a company-intelligence platform built on Thailand's official business registry. He writes about the structure of the Thai economy using primary registry data. His work covers company formation, industry concentration, and regional business activity.
Related insights
This page reports company-level and aggregate statistics only. It names no individuals and makes no claims about any person. Figures are provided for information and may be revised as source data updates.