I’m Trying To Understand Hedonic Adjustments

Brent Donnelly is a senior risk-taker and FX market maker at HSBC New York and has been trading foreign exchange since 1995. He is the author of The Art of Currency Trading (Wiley, 2019) and his latest book, Alpha Trader, hits the shelves in Q2 2021.

You can contact Brent at  [email protected]  and on Twitter at @donnelly_brent.

As with all of our guest contributors, Brent’s post may not represent the views of Epsilon Theory or Second Foundation Partners, and should not be construed as advice to purchase or sell any security.

With the intense sturm und drang around inflation right now, expect the next six months to yield intense hand-wringing and chin scratching over CPI, PCE and … hedonic quality adjustment. Quite often when people go to Twitter to rail about CPI, they say something like “Man, my cost of living doesn’t look like that! My tuition and health care costs are skyrocketing!” Savvy statisticians then respond with a flurry of charts that look like the one at right and say “the plural of anecdote is not data” or something similar.

So all those rising prices are being offset by all those falling prices and people are just noticing the prices that go up, right? Nope. The source of the data on that chart is the BLS, the U.S. organization charged with preparing the CPI data. The data has been sliced and diced like crazy.

The CPI index is not a cost of living measure! It is a hedonically-adjusted basket of assumptions that attempts to track some sort of cost vs. quality / total-utility metric over time. So what exactly is hedonic quality adjustment? Let’s ask the BLS:

Hedonic quality adjustment is one of the techniques the CPI uses to account for changing product quality within some CPI item samples. Hedonic quality adjustment refers to a method of adjusting prices whenever the characteristics of the products included in the CPI change due to innovation or the introduction of completely new products.

The use of the word “hedonic” to describe this technique stems from the word’s Greek origin meaning “of or related to pleasure.” Economists approximate pleasure to the idea of utility – a measure of relative satisfaction from consumption of goods. In price index methodology, hedonic quality adjustment has come to mean the practice of decomposing an item into its constituent characteristics, obtaining estimates of the value of the utility derived from each characteristic, and using those value estimates to adjust prices when the quality of a good changes.

The CPI obtains the value estimates used to adjust prices through the statistical technique known as regression analysis. Hedonic regression models are estimated to determine the value of the utility derived from each of the characteristics that jointly constitute an item.

There is a significant degree of modelling and massaging going on. The BLS dances around the CPI as cost of living measure question here (see question #9) and the media often inaccurately describe CPI data as changes in cost of living. This leads to a situation where people see prices ripping higher and CPI at 2.0% and it makes them want to yell at CPI.

Since I always think of TVs and cars as goods that have seen flat or falling prices over the years, I decided to have a quick peek at the real life vs. BLS-imagined price movements in the world of new vehicles. I made the arbitrary and grossly simplifying decision to use the Honda Accord and the Ford Mustang as my cars because they have been popular for more than 30 years, there is plenty of historical pricing data, and their target demographics and brand image haven’t changed much over time.

Here is the Honda Accord, then and now:

The 1990 vs. 2020 versions are clearly not the same car, especially when you compare airbags (or lack thereof), ABS, computer-assist, and all that stuff. The tricky thing for non-time-traveling humans, though, is that roughly the same socioeconomic cohort that bought Ford Mustangs or Honda Accords in 1990 is in line to buy those same cars in 2021. If you hedonically-adjust away the improvements, you are not talking about cost of living anymore, you are talking about quality-adjusted or utility-adjusted cost of living. And the result is a super-squishy approach that requires a litany of assumptions and leads to model output that is more a vague approximation of some utility-adjusted price, not an index that reflects actual real life changes in price.

The next chart is one I created that shows the evolution of the BLS new vehicle price data (red) along with the average price of a Honda Accord and a Ford Mustang each year. Funny enough, the price of a car and the average hourly wage move up at about the same speed (probably not a coincidence!). That’s how Honda Accords and Ford Mustangs remain workhorse middle class cars for over 30 years. If the price was truly remaining flat as the BLS series implies, we’d all be driving BMWs (and eventually McLarens) as wages go higher in a fairly straight line and car prices remain unch.

Did the price of cars go up, or stay flat since 1990? It’s a surprisingly difficult question to answer!

Nominal prices of cars went up about 150%. Hedonically-adjusted car prices barely moved[1]. Real car prices (adjusted for wage growth) were flat. My brain is becoming a pretzel.

And for good order this analysis is not sensitive to what kind of car you pick. All the lines look about the same, the only difference is the upward gradient of the slope, never the direction.

And when it comes to new vehicles, here are some quality changes that might trigger a hedonic adjustment:

If you’re curious, here is the BLS explainer for hedonic adjustment, which includes this fun stuff:

If the item being modeled is men’s shirts, the independent variables might be sleeve length and fiber composition; a simplified version of a hedonic model for men’s shirts might be:

Here all shirts are either short sleeve or long sleeve and either cotton/poly or 100% cotton. After doing the statistical processing BLS might estimate that ß1 = 0.15 and ß2 = 0.25. This indicates that a long sleeve shirt is 15 percent more valuable than a short sleeve one and that a 100% cotton shirt is 25 percent more valuable than a cotton/poly blend shirt.

If the BLS data collector is forced to replace a short-sleeve cotton/poly shirt in the CPI sample with a long sleeve 100% cotton shirt, the CPI would adjust the price of the old item by the features in the new item, leading to a price adjustment of about 49 percent (e0.15+0.25).

If the price of the original shirt had been $20 and that of the replacement shirt $30, rather than using a $10 increase in price for that sample observation, the CPI would adjust the original shirt for sleeves and cotton content resulting in a price estimate of $29.84 (20*e0.15+0.25). This adjustment attributes most of the price difference between the shirts to the change in characteristics and an increase of only $0.16 is shown.

With 46% of the products in CPI hedonically adjusted, it’s hard to know what CPI is actually telling us. Here is a good description of the problem from the Praxis Advisory blog.

While theoretically attractive, hedonic adjustment misses a key point. In all likelihood the good purchased was for the same or slightly higher price, regardless of quality (have Lexus cars dropped in price over the last 5 years?) Consider the following example. Say the only product that Americans purchase are M&M candies—100 M&Ms in a bag that costs $1.00. Each person is limited to one bag.

Through the miracle of productivity, a way is found to fill each bag with 110 M&Ms that is now priced at $1.10. Hedonic adjustment would say that the bag really only costs $1.00 and that the CPI has not increased, since each M&M still costs a penny each. But the cost of the bag of M&Ms has gone up. And since each person must buy a bag, instead of an individual M&M, their cost of living has gone up by 10%. They must fork over an extra dime even though they’re getting more for their money. We can’t buy individual parts of a new car; we have to buy the whole car, complete with quality improvements. And the whole package costs more, improved or not.

While hedonic adjustment may accurately reflect productivity increases, they don’t accurately reflect America’s cost to live. These adjustments accrue to businesses (which in turn don’t provide adequate raises) and the federal government (which in turn under-compensates Social Security recipients).

Which is all fine and good. But we’re still left with this:

A Honda Accord cost $12,000 in 1990 and it costs $25,000 now.

A Mustang was $9,000 and now it’s $27,000.

The BLS has new car prices close to unchanged over the past 30 years.

Here’s the best my pretzel-brain can do to reconcile this. Here are my takeaways:

  • People think CPI is a cost of living index, but it’s not. Stop comparing CPI to how your cost of living has changed.
  • That said, the BLS does not push back very hard on the idea of CPI as a cost of living index!
  • Even though CPI is not a cost of living index, CPI is still used for cost-of-living adjustments to Social Security and other benefits. CPI is also used to price TIPS.
  • There is a point of abstraction at which the original object is no longer recognizable. It is possible that CPI is such an abstraction and this explains why it does not compute with the everyperson’s day-to-day experience. We are all staring at Salvador Dalí’s “Persistence of Memory” screaming “that’s not a clock!” It IS a clock. But it’s also not a clock.
  • Prices for an item can go up or down and that item’s contribution to CPI can be in the same direction, the opposite direction or zero.
  • Interestingly, hedonic adjustments only act as deflators. Say the airline crams another seat in your row, eliminates carry-on bags and otherwise makes your flight less happy and hedonic. Does the hedonically-adjusted price of your airfare increase? Nope.
  • CPI is a model output dependent on a huge string of assumptions, not necessarily a true reflection of price changes for goods and services purchased by individual consumers.
  • Just because you hear 14 scary inflation anecdotes tomorrow, don’t assume all or any of those price movements will impact CPI in the way that you expect. They might! Or they might not.

This has been my best attempt to better understand the inner workings of CPI and why it looks and smells so different from reality. There is no easy answer to this conundrum as evidenced by this long discussion on the topic on the Social Security dot gov website, but I think this article from Bloomberg on the history of CPI sums it up pretty well:

The first architects of price indexes appreciated the degree to which these numbers are nothing more than vague approximations that, precisely because they rest on such shaky foundations, can be put toward political ends.

Sounds about right. Have a plaid and/or ludicrous day.

[1] I am fairly sure hedonic-adjustment of new vehicle prices started in 1998 / 1999 but I couldn’t find the precise answer.

Brent Donnelly is a senior risk-taker and FX market maker at HSBC New York and has been trading foreign exchange since 1995. He is the author of The Art of Currency Trading (Wiley, 2019) and his latest book, Alpha Trader, hits the shelves in Q2 2021.

You can contact Brent at [email protected] and on Twitter at @donnelly_brent.


U.S. Global Markets Sales and Trading Commentary Disclaimer

This material has been prepared by a member of the sales and trading department of HSBC Securities (USA) Inc. (“HSI”), HSBC Bank USA, N.A. (“HBUS”), and/or its affiliates (individually and collectively, “HSBC”). This material has not been prepared by HSBC’s research department. This material has been approved for publication in the United States by HSI, which is regulated by The Financial Industry Regulatory Authority (“FINRA”) and the Securities Exchange Commission (‘SEC”), and/or HBUS, which is regulated by the Office of the Comptroller of the Currency (“OCC”), the U.S. Commodity Futures Trading Commission (“CFTC”) and the National Futures Association, as a provisionally registered swap dealer. This material is intended for your sole use and is not for general distribution; you may not distribute it further without the consent of HSBC. This material is for informational purposes only and does not constitute an offer or commitment, a solicitation of an offer or commitment to enter into or conclude any transaction or to purchase or sell any financial instrument.
The sales and trading department of HSBC is engaged in selling and trading and may make markets in securities or derivatives to which this material relates or is relevant. Accordingly, recipients of this material should not regard it as an objective or independent explanation of the matters contained herein. Any opinions in this material are the opinions of the author and may be changed at any time without notice. Opinions expressed in this material may differ from the opinions expressed by other divisions of HSBC, including its research department and corresponding research reports. Any information contained in this material is not and should not be regarded as investment research, debt research, or derivatives research for the purposes of the rules of the Financial Conduct Authority, the SEC, FINRA, the CFTC or any other relevant regulatory body. As part of the sales or trading desk, the author may have consulted with the trading desk while preparing this material and such trading desk may have accumulated positions in the financial instruments or related derivatives products that are the subject of this material.
Prior to entering into any proposed transaction, recipients should determine, in consultation with their own investment, legal, tax, regulatory and accounting advisors, the economic risks and merits, as well as the legal, tax, regulatory and accounting characteristics and consequences of the transaction. This material does not constitute investment advice. To the extent permitted by law, HSBC does not accept any liability arising from the use of this communication. This material is for institutional investors and/or institutional accounts as defined in FINRA Rule 4512(c) and 2210(a)(4), or for eligible swap participants as defined in Section 1a(18) of the Commodity Exchange Act.
This material may contain information regarding structured products which involve derivatives. Do not invest in a structured product unless you fully understand and are willing to assume the risks associated with the product. Products described in this material may not be principal protected. Any products stating principal protection apply only if they are held by investor until maturity. If you close out the trade prior to maturity, you may lose your principal.
The securities described herein may not be eligible for sale in all jurisdictions or to certain categories of investors. Unless governing law permits otherwise, the recipient of this document must contact a HSBC affiliate in its home jurisdiction if it wishes to use HSBC services in effecting a transaction in any investment mentioned in this document. The information contained herein is derived from sources believed to be reliable, but not independently verified. HSBC makes no representation or warranty (express or implied) of any nature nor does it accept responsibility of any kind with respect to the completeness or accuracy of any information, projection, representation or warranty (expressed or implied) in, or omitted from, this material. No liability is accepted whatsoever for any direct, indirect or consequential loss arising from the use of this material.
This material does not constitute and should not be construed as a recommendation to enter into a securities or derivatives transaction. If you are considering entering into a security or derivatives transaction or would like additional information, please contact your local sales or trading representative.

To learn more about Epsilon Theory and be notified when we release new content sign up here. You’ll receive an email every week and your information will never be shared with anyone else.


  1. Another hedonic adjustment is how much more pleasurable life is in 2020 than it was in 1990.

  2. CPI is cost of living any time that making it such keeps the government from paying out more of our money to us. All other times it’s definitely not cost of living. Schrodinger’s CPI.

  3. The truth is what we need it to be —-and this , to me, is the biggest example of all. The real inflation rate cannot be known - the Government’s solvency rests on it.

    More and more missionary’s are speaking on the subject and it’s the thing that can bring the whole house of cards down.

    Of course they and their renfields in the media will blame the markets and capitalism for the crash and use the whole crisis to consolidate their power. Our road to serfdom will then be complete.

  4. First question: who benefits from CPI as measured? Second question: who benefits from CPI being under reported? Third question: which societal group has had their share of national income under structural pressure over the post Volker years?

  5. Wouldn’t evenly applied hedonic quality adjustments increase the adjusted ‘cost’ of goods which get worse in quality? I’ve never heard that flow rate limiting shower nozzles which cost the same dollar value as good shower nozzles had a huge upward hedonic ‘cost’ adjustment.

    Ditto for paper-thin jeans, un-flushable toilets, clothes dryers you have to run three times per load, bloatware-encrusted cheap computers, the thing in cars that beeps at you when it’s angry…

    Rather than accept the entire premise and argue with the implementation, I’m going to reject the premise that hedonic quality adjustments are a valid way of measuring constant standard of living. Fortunately, there’s an even simpler reason why we can reject the entire concept. If you assume that overall happiness remains constant across this period of hedonic growth, there’s no reason why we should accept a downwards ‘cost’ adjustment.

    Happiness associated with material goods is largely zero-sum. Humans look to those around them for cues about how valuable our possessions are. It doesn’t really matter to my day-to-day life that I live better than Louis XVI if my neighbor gets a new car and I’m still driving my college POS.

    Furthermore, we still have to pay the prices out of our pocket. Additionally, as one gentleman memorably put it a few years ago, “we can’t eat iPads!”.

    Accepting this simplifies things a lot. Ignore the hedonic adjustments, just do an apples to apples comparison. I’d suggest percent of median income; real income runs into problems if CPI is incorrect. Lets call this new measure subjective inflation and see what it tells us.

  6. I always thought that would make a great thesis for some graduate student.

  7. My God is this just beautifully written and 100% correct.

  8. An under-appreciated issue here is how CPI must be rigged to damp feedback effects. When payments are indexed to inflation metrics (such as social security and union rates), and good/service costs are indexed to inflation expectations via consumer price increases and contractual rate adjustment schemes, you get a self-reinforcing feedback loop. Higher costs → higher income → higher costs. Accurate inflation-measuring combined with automatic contract indexing makes the system dynamically unstable from a control-theory standpoint. It is thus mechanically necessary for system stability that either the inflation adjustments don’t fully compensate for measured inflation, or the inflation metric systematically under-reports real inflation.

    Everybody seems to have forgotten the CPI system was DESIGNED to under-report cost of living inflation.

  9. Avatar for MJ304 MJ304 says:

    Is that after adjustment for COVID? :wink:

  10. Avatar for Pat_W Pat_W says:


  11. Avatar for Pat_W Pat_W says:

    In Latin America all countries I am aware of simply tract a basic “basket of goods” for a family of 4. This includes X quantities of rice, beans, powdered milk, salt, and a few other things. they track the things that matter to people then compare increases in prices month by month and year by year.

    Other commentators pegged the actual intent o the VPI very well. It’s purpose is to keep the U.S government solvent.

  12. To be fair, being old enough to remember my families 1974 Chevy Nova that had its front passenger floorboard rust out to allow direct street ventilation, at less than 60k miles BTW, things do improve and some sort of adjustment is necessary. Does making these adjustments cause other distortions? Sure. But not doing them at all isn’t helpful either. Here is a quote from a recent article, which discusses the “aging American car fleet” which only happens because cars today are much better.

    “In the mid-’90s, 100,000 miles was about all you would get out of a vehicle. Now, at a 100,000 miles a vehicle is just getting broken in,” said IHS Markit’s Todd Campau.


    Another thing to consider. When you dated a girl in the 1970s you knew she was interested in a long term relationship if she had her data “work on the car together”. The dad wanted to know if a future son in law could do basic auto maintenance, which would include changing tires, changing oil and lube, changing spark plugs and point, along with timing the ignition, cleaning and adjusting carbonators, etc. Because if you couldn’t do these things, he knew he would be drafted to do such things. Not only are some of these things not even on cars anymore, spark plug changes occur once or twice in a car’s lifetime. Yes cars cost more, but maintenance is much easier too.

Continue the discussion at the Epsilon Theory Forum


The Latest From Epsilon Theory


This commentary is being provided to you as general information only and should not be taken as investment advice. The opinions expressed in these materials represent the personal views of the author(s). It is not investment research or a research recommendation, as it does not constitute substantive research or analysis. Any action that you take as a result of information contained in this document is ultimately your responsibility. Epsilon Theory will not accept liability for any loss or damage, including without limitation to any loss of profit, which may arise directly or indirectly from use of or reliance on such information. Consult your investment advisor before making any investment decisions. It must be noted, that no one can accurately predict the future of the market with certainty or guarantee future investment performance. Past performance is not a guarantee of future results.

Statements in this communication are forward-looking statements. The forward-looking statements and other views expressed herein are as of the date of this publication. Actual future results or occurrences may differ significantly from those anticipated in any forward-looking statements, and there is no guarantee that any predictions will come to pass. The views expressed herein are subject to change at any time, due to numerous market and other factors. Epsilon Theory disclaims any obligation to update publicly or revise any forward-looking statements or views expressed herein. This information is neither an offer to sell nor a solicitation of any offer to buy any securities. This commentary has been prepared without regard to the individual financial circumstances and objectives of persons who receive it. Epsilon Theory recommends that investors independently evaluate particular investments and strategies, and encourages investors to seek the advice of a financial advisor. The appropriateness of a particular investment or strategy will depend on an investor’s individual circumstances and objectives.