Cambodia: Property Buyer Surveys: Intentions vs Transactions

A buyer survey can be valuable without being a transaction forecast. The key is to read the percentage as an answer to a defined question from a defined group. A location choice is a preference. A plan to buy within a stated window is purchase intention. A view on affordability or future prices is sentiment. Completed sales, title transfers and lending-based price measures are observed market indicators, each with its own coverage. Moving from one category to another requires evidence, not a change of wording.

The 2019 Consumer Sentiment Survey is a useful example because both the headline result and the recruitment setup are visible. Realestate.com.kh reported more than 2,000 responses and asked whether respondents planned to buy property in the next 12 months; about 78% said they did. Read alone, that can sound like a broad estimate of future Cambodian demand. The survey invitation tells a narrower story: participation was open to any site visitor, whether renting, owning or merely considering property, the questionnaire was offered in English and Khmer, and entrants could win hotel packages.

So the defensible statement is that roughly 78% of this survey’s respondents expressed a 12-month purchase intention. It does not establish that 78% of Cambodian households, property seekers or even all visitors to the portal would buy. Nor does it show how many of those respondents later completed a purchase. Turning intention into a transaction forecast would require follow-up on the same respondents or a separately validated conversion model; neither is provided by the public 2019 material used here.

The 2023 survey illustrates the same distinction with a different time horizon. Published results say about one-third of respondents were looking to buy within 18 months and nearly 15% within six months. Those figures help describe the timing of interest among the people who answered. They become evidence about market-wide completed demand only if a genuinely comparable observed series exists for the same period and housing segment. An official number about prices, credit or a different property type cannot fill that gap simply because it comes from a regulator.

Sample size answers only part of the reliability question. More responses can make estimates within a particular respondent pool less noisy, but they do not repair a recruitment process that systematically attracts some people more than others. A property portal, for example, naturally reaches people with an above-average reason to think about property. Two thousand responses from that audience may be highly informative about that audience while still being a poor stand-in for the Cambodian population or for everyone who will transact.

The 2019 material makes this boundary unusually visible. The published overview says 73.4% of the more than 2,000 respondents were Cambodian and 26.6% were foreign nationals, and notes that most respondents had already shown an interest in real estate by browsing Realestate.com.kh. That context changes how the 78% purchase-intention figure should be read. It is evidence of strong stated interest within a self-selecting property-focused audience, not a nationally representative home-buying rate. The survey remains useful; the claim simply has to stay inside the population it actually reached.

For 2022 and 2023, the public result pages provide useful sample totals and many breakdowns, but the accessible methodology is less complete. The 2022 article reports more than 2,300 respondents and says Chroy Changvar was the preferred Phnom Penh location for 21%. The 2023 publication reports more than 2,000 local and foreign respondents and gives figures for budgets, locations, property types and intended purchase timing. In the public material reviewed for this page, however, the full recruitment frame, weighting procedure, non-response treatment and denominator for every published percentage could not all be verified. Those omissions do not invalidate the reported preferences; they limit how far the results can be generalized.

The 2025 academic study is smaller but more explicit about whom it describes. It analysed responses from 410 Phnom Penh residents with stable monthly incomes, using structured questionnaires in English and Khmer and descriptive statistics to examine factors associated with home-purchase intentions and decisions. That makes it a focused study of perceptions among a defined group in the capital. It is not a national buyer census, and it was not designed to count completed property transactions.

Local-versus-foreign composition deserves the same care. A mixed-sample result cannot automatically be applied to a foreign buyer merely because foreigners participated in the survey. A foreign-buyer claim needs the subgroup’s own result, its denominator and enough information about how that subgroup entered the sample. Otherwise the overall percentage can be numerically correct while answering a different question from the one an overseas buyer actually has.

These are survey signals beside the closest available market checks, not forecasts matched to outcomes. Comparability depends on the underlying definitions.

Annual branding does not create a time series. To interpret a change in percentages as a change in buyer sentiment, the underlying measurement needs to remain sufficiently stable: question wording, answer choices, purchase horizon, recruitment route, geography and respondent mix all matter. When those ingredients shift, the percentage can move even if the underlying market has not changed by the same amount.

The available 2019 material refers to an intention to buy within the next 12 months. In 2023, the public results emphasize an 18-month horizon, with a separate figure for people looking to buy within six months. Putting the 2019 figure of about 78% beside the 2023 “about one-third” figure and calling the difference a collapse in demand would therefore be misleading. The time window is different, and the accessible 2023 results page does not disclose the full questionnaire wording and denominator with enough detail to make a clean like-for-like calculation.

The 2022 publication demonstrates another kind of mismatch. It reports more than 2,300 respondents and a location-preference result: Chroy Changvar rose from 12% in 2017 to 21% in 2022 among respondents to the relevant surveys. That is interesting evidence about stated location choice. It is not evidence that transaction volume in Chroy Changvar rose by a comparable amount. Without the full questionnaire and a confirmed stable recruitment frame, the safest interpretation is a preference signal from that edition of the survey rather than a continuous market-demand index.

Cross-year comparison is still possible, but the level of certainty has to match the documentation. When wording and respondent bases are demonstrably comparable, percentage changes can describe a trend in survey answers. When only the subject matter is similar, it is more defensible to compare broad patterns—such as which locations or property types attracted attention—without treating percentage-point movements as precise market changes. If a methodological field is not published, “unknown” is a more accurate entry than an inferred answer.

The useful market check is the one that most closely matches what the survey claimed. A purchase-intention question calls for observed sales activity over a similar horizon and, ideally, the same property segment. A price index answers a different question. Mortgage observations answer another. The goal is not to find an official statistic that makes the survey look right or wrong; it is to see whether there is an observed measure close enough to test the same proposition.

Cambodia’s National Bank reports show why property type cannot be ignored. For 2022, the bank estimated that sales of houses—including flats, twin villas and single villas—increased by about 80%, while condominium sales fell 58.2%. In 2023, the directions diverged again: sales of the house group fell 49.7%, while condominium sales increased 2%. The 2024 report then recorded condominium sales up 35.5% and house sales down 12.5%. A single survey percentage labelled as “buyer demand” would hide these very different segment paths.

Those official figures do not prove that the 2023 survey was inaccurate. The survey’s published purchase-intention figure covered an up-to-18-month horizon, included local and foreign respondents, and was not limited to one housing type. The NBC indicators describe sales by their own categories and periods. The most that can be said is that the survey signal and later observed activity are only partially comparable: positive stated intention could coexist with a sharp contraction in one segment and expansion in another. There is no defensible intention-to-purchase conversion rate in these sources.

The Residential Property Price Index needs an additional distinction. The IMF’s 2024 technical report explains that the NBC index is compiled from residential property loan data submitted by mortgage-lending institutions and adjusted for differences in property characteristics. Its coverage is limited to the portion of the residential market financed by commercial-bank loans. Cash purchases, developer-financed transactions and purchases financed through foreign banks sit outside that sample, and the report notes particularly few observations for condominiums and apartments. That makes the RPPI useful for tracking prices within its covered market, not for counting all property deals.

As of 29 September 2026, the official public sources reviewed for this page did not yield an open, methodologically documented time series of registered ownership transfers that can be aligned directly with the 2019–2023 survey windows. That absence matters. NBC annual reports provide sales-direction estimates, while the RPPI is a lending-based price measure; neither should be relabelled as a registry of completed title transfers. A missing direct comparator is a legitimate research result, not a reason to substitute the nearest official series.

For a reader, the practical test is therefore four questions. What exactly did the survey ask? Who was eligible or likely to answer? What period did the answer refer to? Which observed measure covers roughly the same period and property type? If the fourth answer is “none in open data,” the survey can still inform expectations and preferences, but the claim must stop short of completed transaction volume.