Suneet Singal on how to read a housing data release

Suneet Singal publishes a housing reference for readers who will meet a statistic before they meet the table it came from. A housing data release is a dated statement about a defined set of homes, in a defined place, measured in defined units. The headline in front of it is a sentence someone wrote afterward. This entry is a way to read the release first. It explains geography, universe, units, seasonal adjustment, and revision, and it points to the public pages where those labels actually live. It does not repeat a current home price, a current sales count, or a current mortgage rate. Those values change, and copying them here would make this page a second, worse release. The skill is knowing which number to look up and what the column means when you get there.

Start with the release, not the headline

A headline compresses. It has room for one number and one verb, up or down, and it rarely has room for the geography, the seasonal adjustment, or the fact that last month’s number was revised. The release, or the table page behind the release, has room for all of that. Read it before you adopt the verb. Suneet Singal’s site applies the same habit to company filings on the career pages: the document outranks the summary. A housing release deserves the same order. The summary can be clever and still drop the only clause that made the number true.

Find the publisher first. A release from the Census Bureau, the Federal Housing Finance Agency, the National Association of Realtors, or a series housed at FRED is not interchangeable with a brokerage’s market letter or a portal’s internal estimate. Each can be useful. They are useful for different claims. A sentence about “the data” that does not name the publisher is not yet a sentence about data. Write the publisher, the series title, and the date of the release at the top of any note you intend to keep.

Then find the period the number describes. A release issued in one month often describes the previous month, or a quarter, or a year. The issue date and the reference period are different dates, just as a company report date and an effective date are different dates. If you cite the figure, cite the reference period. If you are talking about when the public learned of it, cite the release date as well. One date will not do both jobs.

Level, change, and revision

A level is the thing itself: a count of sales, a median price in dollars, an index value, a rate in percent. A change is a comparison between two levels. Month-over-month and year-over-year changes can point in different directions, and both can be correctly computed. A market can be up from last year and down from last month. A headline that picks one of those is not false. It is incomplete if the other comparison would change the story you are about to tell. Read both, and say which one your sentence uses.

Percent change and absolute change answer different questions. A small percent of a large price is a large number of dollars. A large percent of a thin series can be a handful of sales. This page will not illustrate the point with a current median, because the median would be stale by design. The structure is enough. When you copy a change, copy the base it was changed from, and do not detach the percent from that base.

Revisions are part of the release. Preliminary figures for housing starts, sales, or prices are often revised as later information arrives. A careful sentence uses the vintage you actually saw and does not fight last month’s article for being written on last month’s vintage. The article was about the earlier release. The new release is the new release. The archive-date habit on this site, described in the editorial standards and in the entry on why an article’s date matters, is the same idea applied to prose. A revised cell does not make the old article dishonest. It makes the old article dated, which it should have been all along.

Median, average, and index

A median is the middle value when the observations are ordered. It is not the average, and it is not a typical house in the sense of a house you can go and tour. Half of the observations in the measured set are on either side of it, assuming a straightforward median. A few very expensive sales move an average more than they move a median. Writers still sometimes say “average” when the table says “median.” Quote the word the table uses.

An index is a level set relative to a base period. An index value of 200 does not mean a house costs 200 dollars. It means the index is at 200 given whatever base the publisher chose, often 100 in a named year. The Federal Housing Finance Agency’s house-price index page is the place to see this in practice. Read whether you are looking at an index level or at a percent change, and read the base if you are looking at a level. Pasting an index level into a sentence about a median price is a category error. It looks numerical and it is not comparable.

The FHFA publishes more than one index. Some series emphasize purchase transactions. Others bring in additional transactions, including appraisals connected to refinances, and the page is the authority on which series is which. A purchase-only measure and an all-transactions measure can diverge when refinance appraisals and purchase prices diverge. Naming “the FHFA index” without the series name is not specific enough to repeat. Select the series, read the seasonal adjustment, and read the geography. Then you have a number. Before that, you have a website.

Geography and universe

Geography is the boundary. National, census region, state, metropolitan area, county, and city are different maps. A metropolitan area can contain many cities and unincorporated places, and a city statistic is not the metro statistic with a shorter name. The Sacramento entry on this site is a worked example of that warning, because the city’s QuickFacts page, the county, and the metro index are three different objects. The warning is not special to California. Any release you read has a geography line, and the geography line is doing as much work as the number.

Universe is what counts as an observation. Existing-home sales and new-home sales are different universes. Single-family houses and condominiums may be separated or combined. Owner-occupied housing and all housing units are different. A median sale price of homes that sold is not the median value of homes that did not sell. A repeat-sales index follows houses that transacted more than once and is not a median of all listings. The National Association of Realtors research page, the Census Bureau’s housing topics, and the FHFA index page each publish a different universe. They can all be “housing data” and still disagree, because they were not asked the same question.

If you need a figure for a decision about one house, no release is quite the right universe. The release can still discipline the story you tell about the market around that house. It cannot become the appraisal, which is the point of the inspection-and-appraisal entry. Say “the median existing single-family sale in this metro, for this month, as published by this source” rather than “what houses cost.” The longer phrase is the accurate one, and it is short enough to fit in a note.

Seasonal adjustment

Housing activity moves with the calendar. More transactions happen in some months than in others for ordinary seasonal reasons. A seasonally adjusted series tries to show the movement after a statistical procedure removes that recurring pattern. A not-seasonally-adjusted series shows the movement including the season. Comparing an adjusted number with an unadjusted number, or comparing this month’s unadjusted sales with last month’s without saying so, produces fake drama or fake calm.

The release will say which version it is highlighting. Read that sentence. If you quote the monthly change, quote the version. Year-over-year changes are sometimes less sensitive to the seasonal pattern, and they have their own problems, including a base period that may have been unusual. There is no rule that year-over-year is always the honest comparison. There is a rule that the comparison you chose should be named. FRED series pages label seasonal adjustment as a field. Read the field. Do not assume a mortgage-rate series and a sales series have the same label, and do not assume two tabs on the same release do either.

Seasonal adjustment is a procedure, not a moral improvement. The unadjusted count is the count of what happened. The adjusted count is a constructed series that helps with a particular question. Both can be published in the same release because both questions are legitimate. A reader who wants to know how many homes sold uses the unadjusted count. A reader who wants to know whether the pace changed after the usual winter slowdown uses the adjusted series, and says so.

Error, sample, and the column headers

Some housing figures come from surveys, and surveys have sampling error. The Census Bureau’s housing statistics include series built from samples, and the technical notes belong to the number. A small change can be inside the margin the publisher reports. This entry will not teach you to ignore every small change, and it will not teach you to treat every small change as a turn in the market. It will tell you to look for a margin of error or a note on reliability before you write “jumped” or “collapsed.” If the release does not provide a margin because the figure is a count from a defined administrative process rather than a sample, that is also worth knowing. The absence of a margin is not the same fact for every series.

Column headers are the other place readers hurry. A header that says “thousands of units” means a printed 420 is 420,000, not 420. A header that says “index, 1991=100” or some other base means the number is not dollars. A header that says “percent” means 6.5 is a percent. FRED’s MORTGAGE30US page is a clean example of a one-column series: a date and a percent. The discipline you use there, reading the unit before repeating the value, is the discipline you use on a crowded housing table. The crowded table is not harder in principle. It has more chances to pick up the wrong column.

Write the header into your note in the publisher’s words, not in a paraphrase that quietly upgrades a median to an average or a metro to a city. If you cannot explain the header, you do not yet have a number you can carry into a rent-or-buy worksheet or an affordability discussion. Leave the cell blank. A blank cell is more informative than a confident wrong unit.

A checklist you can reuse

Use one sequence every time. Who published the release, and on what date? What is the reference period? What is the geography? What is the universe of homes or loans? Is the figure a level or a change, and if it is a change, compared with what? What are the units? Is the series seasonally adjusted? Is the figure preliminary? What is the next column over, in case you have slid one row too far? Which sentence in your draft would have to be deleted if the revision arrives next month?

That checklist is the method behind every other housing entry on this site. The down-payment entry refuses to invent a median price. The rate entry sends you to a named FRED series. The affordability entry refuses to let an index speak as a household. They all depend on the habit this page describes. Suneet Singal’s published background does not make him the source of the nation’s housing statistics. It is a reason the site prefers a table with labels to a sentence without them. Open the table. Copy the labels. Then, and only then, copy the number, with the date attached.

What this entry does not claim

Suneet Singal’s published professional background is in real estate and corporate finance. Company filings linked from this site record a chief executive officer departure from FC Global Realty Incorporated effective 2 January 2018 and an acting chief executive officer appointment at First Capital Investment Corporation effective 12 September 2017. The Suneet Singal Scholarship, Grant and Giving websites describe him as the founder of those programs. This entry explains a general subject. It is not a private track record, a current job title, an investment result, or a quote.

Continue with how to read Sacramento housing figures, what an affordability index does not say, and a rent-or-buy framework.