The Macro Code #9 – If you can't raise prices, then you may lower the quality
Could the measurement of Quality-Adjusted Price Changes have any flaws?
Welcome to The Macro Code, an open diary on markets, macro-trends and structural changes.
The debate over transitory vs. permanent inflation seems to be over as almost all consensus economists (with a few exceptions) have begun to warn of the current inflationary pressure. Price levels are rising sharply in many sectors. These continued signs of rising inflation are beginning to worry monetary policymakers who may be prompted to intervene more quickly and to a greater extent than is currently being discounted by the markets.
However, despite inflationary pressures, some companies haven’t been raising prices. Instead, they’ve been cutting back customer services and conveniences. As a result, rather than increasing the price of a good or service in order to keep up with inflation, some companies prefer to drop the quality of the good/service offered. In this way, the inflationary impact on the consumer may prove to be less noticeable.
In this weekly newsletter, we examine the dynamics of shadow inflation: we look at how inflation measurement adjusts for changes in quality.
Our research question is the following: are pandemic-induced quality degradations on goods and services fully reflected in economic statistics?
If policymakers and investors do not sufficiently account for the adjustment for quality in their estimates of inflation levels, the risk is that they will misjudge the phenomenon.
Evidence of lower quality is widespread across sectors
Recently, a NYTimes article highlighted how some companies facing supply disruptions and labor shortages have addressed these issues not by raising prices (or not just raising prices), but by taking steps that might give their customers a lesser experience. Customer sentiment on restaurant cleanliness fell 4.2 percent this year, according to Black Box Intelligence, which tracks online reviews of 60,000 restaurants. Complaints have been frequent about the cleanliness of tables, floors and bathrooms. Satisfaction with customer service was also down, especially regarding beverages, with guests complaining more about receiving the wrong order or no drink at all.
People trying to buy appliances and other retail goods are waiting longer. According to J.D. Power, even at the highest-rated retailers, only 57 percent of customers were able to get customer service within five minutes this year, down from 68 percent in 2018.
Three examples related to our daily lives may be useful to better understand the phenomenon.
Gym membership: If you have joined a gym in the last 6 months, you will have noticed that you have had to pay a full price membership in most cases, with limitations on the services available (closure of relaxation areas, sauna, removal of shared phones in the locker rooms, etc.). The price of the subscription has remained the same as in the pre-pandemic era, the quality of service has decreased.
Hotel service: if you stayed in a hotel or a B&B last summer, you will have noticed that in many cases the cleaning service of the rooms was limited compared to the usual in order to limit opportunities for contagion. Also in this case, the average overnight rate remains the same as in the pre-pandemic era, but with a lower quality of service.
Enrollment in a college course: you may have noticed that tuition fees for a college course, such as a master's degree program, have not dropped from pre-pandemic levels. Yet, most classes in the past 18 months have been held online and paying students have not been able to take full advantage of on-campus university space and all the facilities that are regularly made available to freshmen.
Government statistical agencies try to take changes in product quality into account when calculating inflation. But this process, known as hedonic adjustment, applies relatively easily to physical objects, while it is more difficult with respect to services. For example, it is relatively easy to estimate the value of the quality of the stitching on a pair of pants or the functionality of a remote control for a new model of television. But quality changes that affect customer service can be ambiguous and difficult to measure.
The measurement of quality-adjusted price changes
The empirical literature is in agreement that one of the major source of bias in the measurement of inflation is held to be its inability to properly incorporate quality changes. This is not to say statistical offices are unaware of the problem. Price collectors attempt to match the prices of “like with like” to minimize such bias. However, comparable items are often unavailable, and methods of implicit and explicit quality adjustment are not always considered satisfactory.
As a starting point, it should be specified that a key problem for the goods and services in the CPI sample is that their characteristics, not just their prices, change over time as retailers introduce new versions of items and discontinue previous versions. In many item categories, this is the primary time when a price change occurs. The new version of the item may provide additional benefits or, in some cases, reduced benefits. This change in benefit is the change in quality.
To accurately measure the price change, the CPI must be able to distinguish the portion of the price change due to this quality change. The traditional CPI solution to this problem is to temporarily remove an item from the sample when its quality has changed. Although this method is sometimes acceptable, it affects the CPI if the changes in the price of the new version are systematically different from the changes in the price of unchanged goods.
How to account for product quality change within some samples of CPI items?
The main approach is the use of hedonic regressions in which the price of a model, for example, of a personal computer is regressed on its characteristics. Therefore, hedonic quality adjustment refers to a method of adjusting prices whenever the characteristics of products included in the ICC change due to innovation or the introduction of completely new products.
More technically speaking, a hedonic regression is estimated that includes the characteristics of the variety and dummy variables on time, the coefficients on the time dummies being estimates of the changes in price having controlled for changes in characteristics. This is referred to as the time dummy variable hedonic method. The quality-adjusted price index is taken from the coefficients on the time dummies in the hedonic regression. There is usually little by way of data on quantities, and thus weights, in these estimates. Yet estimates from hedonic regressions have been used to benchmark the extent of bias due to quality changes in consumer price indexes.
There are of course other variants of the time dummy variable method. A sales-weighted least squares estimator could be used, or estimates could be made on a chained basis with, for example, a comparison between January and February being based on hedonic regressions for these months only, with a time dummy for February, and similarly for February and March, March and April, and so on. The estimates of price changes over these binary comparisons would be linked by successive multiplication to form a chained estimate over the whole period.
However, the government's ability to adjust for quality is inherently limited. Statistical agencies have limited budgets, a limited set of objective measurements, and a limited understanding of consumer preferences over time.
The Bureau of Labor Statistics, which generates the consumer price index, does not incorporate quality adjustment on 237 of the 273 components that go into the index, including the vast majority of services (see here).
In normal times, this introduces a little pessimistic bias: statisticians miss some quality improvements and, as a result, official figures tend to overstate the rate of inflation and underestimate the increase in living standards. Thus, as FullStack Economics points out, official measures of inflation in ordinary times may have overstated inflation because there were many under-the-radar product improvements not captured by the data.
Now we're going in the opposite direction: goods and services are getting worse faster than the official statistics admit, which suggests that our inflation problem may actually be bigger than the official statistics suggest.
It's a matter of adjustment...
The basic problem is that economic statistics do not adjust for the quality factor. In normal times, when a new product replaces an old one, price changes might reflect differences in quality rather than customers getting more or less for their money. Therefore, agencies like the BLS need quality adjustments to calculate the true rate of inflation. This is more complicated than it sounds, as we have seen previously there is a big difference between including quality change in the calculation models in the goods sector versus the services sector.
This is precisely why in Chapter 5 of the Boskin Report, a very substantial study of inflation measurement commissioned by the Senate Finance Committee in the 1990s, the committee argued that the BLS did not fully account for quality improvements, leading to a small but consistent overstatement of the inflation rate.
Over the past 18 months, however, we have seen the same problem, but in the opposite direction: many products are deteriorating in quality because of the pandemic. But the BLS does not have the manpower, or in some cases, even the conceptual framework, to fully capture all of these changes in its inflation measurements.
Of course, it should be specified that by this reasoning we are not trying to argue that companies, firms, or businesses have intentionally reduced quality in order to hide the inflationary effect on prices.
There are two scenarios:
i) sometimes, product degradation comes with a plausible COVID-19 justification, but it also conveniently reduces labor costs.
ii) sometimes, the problems are clearly downstream of the pandemic difficulties, but have no obvious justification for consumer welfare: e.g., longer shipping delays.
This reasoning does not go against the work of entrepreneurs. All sectors of the economy have faced great difficulty in overcoming the pandemic. Everyone has done their best to stay afloat, however, there is no denying that some of the pre-2020 product quality has clearly been lost and these losses are not always reflected in the official data.
Additional readings:
The Measurement of Quality-Adjusted Price Changes - National Bureau of Economic Research
Adjusting for quality change - Consumer Price Index Manual
The Macro Code is an open diary on markets, macro-trends and structural changes.
To receive updates in newsletter format, follow the form below.
Please, note that The Macro Code represents personal views only.