Market study for a property development in Morocco: the method, source by source

Everything starts with three framing decisions. If they are wrong, the rest is wrong too, however good the data.

• The product. Villa plots, plots for apartment buildings, apartments, ground-floor retail: different buyers, different prices per square metre. Each product is studied separately, then compared.
• The competitive catchment. We work in widening radii around the site. The first ring holds the properties your buyer will visit on the same day as yours. The outer rings confirm the trend and fill the sample when the first ring is too thin.
• The question. Exit price, sales pace, positioning against a competitor, feasibility of a land purchase: each question calls for a different reading of the same data. It is written at the top of the report.

Framing also fixes the study date. Every price is brought to that date, and it appears on every table.

No single source is enough. Each one measures something different and has its own flaws. A defensible price comes from the way they cross-check.

Sales registered at the land registry (ANCFCC). These are prices that were actually signed, attached to a land title, and therefore to a parcel that can be placed on the cadastral plan. It is the strongest source: it shows what was paid, not what is hoped for. Its limits are well known. The price is the one declared in the deed. And the available database may stop a year or more before the study, which makes indexation necessary.

Online listings on property portals. They are numerous, recent, and describe the product well: area, floor, specification, photos. But they are asking prices, not sale prices. They are located by district, rarely by address. The same listing may appear several times. We therefore apply a negotiation discount, written in the report and presented for what it is: a professional convention, not a measurement.

The price reference of the Direction générale des impôts (DGI). It gives a benchmark by zone and property type. We cite it with its edition. It works as a consistency check: if the first two sources sit far from it, the gap has to be understood. It is not a market price and we never present it as one.

The three sources stay separate to the end. Every table in the report carries the name of its source, so that an asking price is never read as a signed price.

A raw sales database contains transactions that say nothing about the market. Before any calculation, we remove:

• sales of undivided shares: a share trades at a discount, so its price does not reflect the whole property;
• auction sales: the price results from a procedure, not from free negotiation;
• partial sales: a fraction of a parcel or property, where the area actually sold is uncertain;
• transfers at administered prices: the price is set by a schedule or an agreement, not by the market.

That leaves the outliers: a data entry error, a wrong area, a sale between relatives. To remove them without personal judgement, we apply Tukey's rule. It takes three moves. Sort the prices per square metre from lowest to highest. Find the first quartile (a quarter of sales are below it) and the third quartile (a quarter are above it). The gap between the two is the interquartile range. Any sale more than 1.5 times that range below the first quartile, or above the third, is removed.

The rule is mechanical: it does not depend on the analyst's opinion. Above all, the report states how many rows were removed, at each step and for which reason. A starting sample and a final sample with nothing in between cannot be checked.

On the cleaned sample, we work with the median and quartiles, not the mean. The median is the price that splits the sales into two equal halves. One abnormal sale shifts a mean; it barely moves a median. The quartiles give the range that holds the middle half of the market, which is what later lets you place your product at the bottom, the middle or the top.

A sale from three years ago and a sale from last quarter cannot be compared as they stand. To bring them to the study date we use the real estate asset price index (IPAI)of Bank Al-Maghrib, Morocco's central bank, which is built from sales registered at the land registry.

Indexation is done sale by sale, not in bulk. For each row we take the index for the relevant city and property category, at the sale date and at the study date; the ratio between the two is applied to the signed price. An apartment is not indexed with the land index, and a sale in one city is not indexed with another city's series.

The workbook we deliver keeps both columns: the original price and the indexed price, with the index used. You can redo the calculation row by row.

The limit has to be stated: an index describes a trend for a city and a category, not for a street. When a district has moved faster or slower than its city, indexation does not capture it. That is one more reason to cross-check with recent listings.

A price without a location is worth little. Each sale we keep is tied to its land title and placed on the cadastral plan, then overlaid on a satellite image. You then see what the sale really concerned: bare or built land, on an avenue or at the end of a cul-de-sac, inside a completed subdivision or on the outskirts.

The reading then proceeds by widening radii around the site. The radii are not constants: they are set according to how dense the area is and how many sales are available, and they are written in the report. In a dense district the first ring is often enough. On the outskirts the catchment has to be widened, and that has a cost: a wider sample is less comparable, and the report says so.

Listings can only be placed by district. They appear on the map at that scale, with a different symbol, so as not to suggest a precision they do not have.

Past sales show where the market was. Competition shows who you will be selling against. We survey the developments and subdivisions on sale within the catchment, and for each one record:

• the product offered and the areas;
• the list prices, noted on a precise date;
• the specification: finishes, shared amenities, parking, servicing for a subdivision;
• progress: earthworks, structure, delivered;
• remaining stock, when it can be observed.

This survey has limits. A price displayed in a sales office is still an asking price. Remaining stock is not always disclosed, and when it is, it cannot be verified. The report separates what was observed on site from what a sales agent stated.

The competition survey also helps estimate how fast the market absorbs supply, a question our method treats separately.

The next step turns three readings into one reference price per product. The rule is simple: the weights are written down, and registered sales weigh the most, because they are the only prices actually paid. The weights are not constants. They depend on the quality of each source for the project: a large, recent sales sample weighs more than a thin, old one.

The point of this layout: if you disagree with a weight, you change it and see the effect on the result. The price is not an assertion; it is a calculation that can be discussed.

The reference price is not yet the price of your product. The spreads inside the sample still have to be read: the effect of size (small units often sell for more per square metre than large ones), frontage, floor level, specification. These spreads are measured on the data when the sample allows it, and flagged as a judgement when it does not. That reading then feeds the pricing of plots and units, plot by plot or apartment by apartment.

• The report: the question, the catchment, the sources, the cleaning steps with the number of rows removed, the reference prices per product, the limits.
• Maps of the comparables on a satellite image, showing the site, the radii and every sale used.
• The data workbook, row by row: every sale, every listing, the original price, the indexed price, and the reason for exclusion where relevant.
• Planning appendices: zoning, applicable rules and the projection of the site onto the development plan (plan d'aménagement), to check that the product studied is one that may actually be built.

The report follows the principles of the RICS Red Book and IVS standards: written assumptions, named sources, declared limits. It is documented and can be checked row by row; it binds no one, and the workbook is what lets a third party audit it. The assignment is quoted on request, according to scope.

• The data is dated. The sales database may stop before the study date. Indexation corrects for the general trend, not for a recent local reversal.
• Prices are declared. We work with the price written in the deed. We can neither verify nor correct it; we can only cross-check it against the other sources.
• Listings are not verified. Area, condition, price: all of it is self-declared, and the negotiation discount remains a professional convention.
• The land market is thin. For a large site or a rare product, comparable sales can sometimes be counted on one hand. The study says so, widens the catchment, and hands over to another approach.
• A study describes a market; it does not promise sales. The reference price is what the market has accepted for comparable properties. Commercial success also depends on the product, the timing and how the sale is run.

A market study is not an end in itself. It provides the exit prices and the reading of competition. Pricing then distributes that average price across plots or apartments: the method is detailed in our article on the plot-by-plot pricing grid for a subdivision.

Finally, exit prices go into the development appraisal, which derives the maximum price payable for the land. This reasoning, the residual method, is explained in our article on residual land valuation for developers in Morocco, and you can test orders of magnitude with the developer feasibility calculator. An appraisal is only as good as its exit prices, which is why the study comes first.