Knowledge base / Mechanisms
CPI Manipulation and the Measurement of Inflation
The Consumer Price Index is the single most cited number in modern macroeconomics. It sets cost of living adjustments on Social Security, indexes tax brackets, anchors central bank policy, deflates GDP, and frames every news story about prices. It is also a constructed number, governed by methodology choices that have changed substantially since 1980, and almost every methodology change has reduced reported inflation relative to what the previous method would have produced. This file walks through what CPI is supposed to measure, what it actually measures today, who changed it and why, and what the cumulative gap looks like when checked against money supply growth, asset prices, sector specific inflation, and lived household experience.
The argument is not that the Bureau of Labor Statistics is fabricating data. The BLS publishes its methodology in detail. The argument is that the methodology itself, taken at face value, no longer measures the cost of maintaining a fixed standard of living. It measures something narrower, and the narrowing has been politically convenient for the Treasury, for the Federal Reserve, and for any government issuing inflation indexed liabilities.
What CPI Is Supposed to Measure
The original conception of a consumer price index is a fixed basket of goods and services purchased by a representative urban household, priced at two points in time. Divide the cost of the basket today by the cost of the same basket in the base period and you get an index of how much more or less it costs to live the same life. This is the Laspeyres index, and it was the operating principle of the US CPI from its introduction in 1919 through to the late 1990s [1].
The BLS describes the modern CPI as “a measure of the average change over time in the prices paid by urban consumers for a market basket of consumer goods and services” [1]. The phrase “market basket” still appears in BLS documentation. In practice, what is inside that basket, how the items are weighted, how quality changes are handled, and how housing is priced have all changed in ways that bias the result downward.
A pure fixed basket index is unbiased in one specific sense: it answers the question “how much more does it cost to buy the same things this year as last year?” That is the question a retiree on a fixed income, a wage earner negotiating a raise, or a saver trying to preserve purchasing power actually wants answered. It is not the question modern CPI answers.
Pre-1980 Methodology
Before the 1983 housing change and before the 1996 to 1999 wave of revisions, US CPI looked considerably more like a household cost of living index than a substitution adjusted consumption index.
Two features of the pre-1980 methodology matter most:
-
Fixed weights. The basket weights were updated periodically (every ten years or so) and held constant in between. If beef became more expensive relative to chicken, CPI captured the price increase. It did not silently rebalance the basket toward chicken.
-
Housing measured by actual home purchase prices and mortgage interest. Until 1983, the housing component of CPI used the asset price of homes (purchase price) and the financing cost (mortgage interest rate) as direct inputs [2]. When mortgage rates rose from 7 percent in 1976 to 18 percent in 1981, CPI registered that as a massive increase in the cost of housing, because for any household actually buying a home in 1981 it was a massive increase.
The combination produced a CPI that tracked household experience reasonably faithfully. It also produced the headline inflation prints of the late 1970s (13.5 percent in 1980 [3]) that made indexed Social Security expensive and made nominal interest rate caps on bank deposits politically untenable. The pre-1980 CPI was honest enough to be politically dangerous.
The 1983 Housing Change: From Home Prices to Owners Equivalent Rent
In January 1983, the BLS replaced the home purchase and mortgage interest method with Owners Equivalent Rent (OER) for the homeowner housing component [4]. OER is calculated by surveying homeowners and asking, in effect, “what would your home rent for if you were renting it out?” The answer is then used as the price input for the housing component of CPI.
Housing is the largest single category in CPI by weight. As of 2024, shelter is roughly 36 percent of CPI-U, and OER alone is about 26 percent of the index [5]. A change to how housing is priced is a change to roughly a third of the entire index.
The effect since the early 1990s is large and easy to demonstrate. The S&P CoreLogic Case Shiller National Home Price Index has risen approximately sixfold from January 1995 to early 2024 [6]. The CPI Owners Equivalent Rent series has risen approximately 2.5 times over the same window [7]. Both numbers are published. The gap between them is the size of the methodology choice.
| Period 1995 to 2024 | Index growth |
|---|---|
| Case Shiller US National Home Price Index | ~6.0x [6] |
| BLS Owners Equivalent Rent of Residences | ~2.5x [7] |
| BLS Rent of Primary Residence | ~2.4x [8] |
| Headline CPI-U | ~2.1x [3] |
Both views can be defended. OER tries to measure the consumption value of owning a home, on the theory that homeowners are not “consuming” the asset price, only the housing services. The opposing view is that for any household trying to enter homeownership for the first time, the asset price IS the cost of housing, and removing it from CPI removes the largest single inflation transmission channel from the headline number.
The 1983 change is the foundation on which every later methodology revision sits. It is also the change that mechanically breaks the link between CPI and asset price inflation, which is exactly where post-1971 fiat money expansion has shown up most forcefully.
The Boskin Commission, 1996
In 1995 the Senate Finance Committee appointed an Advisory Commission to Study the Consumer Price Index, chaired by Stanford economist Michael Boskin. The commission’s final report, Toward a More Accurate Measure of the Cost of Living, was delivered in December 1996 [9].
The report’s central finding was that CPI overstated the true increase in the cost of living by approximately 1.1 percentage points per year (with a plausible range of 0.8 to 1.6) [9]. The commission identified four sources of alleged overstatement:
- Substitution bias (consumers move to cheaper alternatives within and across categories).
- Outlet substitution bias (consumers move to discount retailers).
- Quality change bias (improvements in products are recorded as price increases).
- New product bias (new goods enter the basket too late, after their initial price drops).
The commission recommended addressing all four through methodology changes, most of which were subsequently implemented by the BLS between 1996 and 1999.
The political context is documented in the report itself. The opening sections discuss the fiscal implications openly. Social Security cost of living adjustments, federal civilian and military retirement, food stamps, veterans benefits, and tax bracket indexing were all linked to CPI. Reducing CPI by 1.1 percentage points per year was estimated to reduce the federal deficit by over $1 trillion cumulatively over twelve years [9]. The report argued this was a side effect of more accurate measurement, not the goal. Critics argued the goal was selected first and the methodology to achieve it was reverse engineered.
Whichever framing one prefers, the timeline is clear: a Senate appointed commission identified a budgetary problem caused by a statistical series, recommended methodology changes that would reduce that series, and the BLS implemented those changes. There is no reasonable reading of the historical record in which the Boskin Commission was a politically neutral measurement exercise.
Substitution Bias Adjustment, 1999
In August 2002, the BLS introduced the Chained CPI-U (C-CPI-U) [10], and starting in 1999 incorporated geometric mean aggregation within most basic CPI item categories [11]. Both changes are forms of substitution adjustment.
The intuition is straightforward. If beef rises 20 percent and chicken stays flat, consumers buy less beef and more chicken. A pure Laspeyres fixed basket index records the full beef increase weighted by the original beef share of spending. A substitution adjusted index allows the weights to shift toward chicken, recording a smaller overall increase.
Two arguments cut against each other:
For substitution adjustment. It tracks what consumers actually buy. If nobody is buying steak any more because steak is too expensive, the price of steak in their actual basket is irrelevant.
Against substitution adjustment. It measures the cost of “the same standard of living however degraded.” If a household used to eat steak twice a week and now eats chicken twice a week because steak became unaffordable, the household has experienced a real reduction in standard of living. A cost of living index that records “no change” is misleading by construction. It is measuring household adaptation, not the cost of living the same life.
The Chained CPI runs roughly 0.25 to 0.3 percentage points per year below CPI-U on average [10]. Geometric mean aggregation within categories adds approximately another 0.2 to 0.25 percentage points per year of downward adjustment versus the previous arithmetic method [11].
Compounded over 25 years, a 0.5 percentage point per year downward adjustment compounds to roughly 13 percent lower cumulative reported inflation. Compounded over 45 years (1980 to 2025) it is roughly 25 percent lower.
Hedonic Adjustments
Hedonic regression is the technique the BLS uses to separate “pure” price change from quality change. The BLS publishes its hedonic methodology by item category [12]. The principle: if a 2024 laptop is twice as fast as a 2014 laptop at the same nominal price, the laptop has gotten cheaper per unit of computing power. Hedonic adjustment records this as a price decrease for the “computing” component of the basket.
Two technical objections are well documented in the academic literature:
-
Selective application across categories. Hedonic adjustment is aggressively applied to electronics, appliances, vehicles, and other categories where measurable quality has risen. It is not symmetrically applied to housing (a 2024 house is in many cases the same physical structure as a 1990 house, but the price has risen 4x to 6x), to healthcare (same procedures, same drugs, much higher prices), or to education (same lecture halls, same degrees, much higher tuition). Categories where prices rise less get a downward adjustment for quality. Categories where prices rise more do not get a corresponding upward adjustment for “quality stagnation” or “deterioration” [13].
-
Quality versus capability. A 2024 car has airbags, ABS, and emissions controls a 1990 car did not have. Most of these are mandated by regulation, not optional features the consumer chose to buy. Treating regulatory mandated equipment as a “quality improvement” the consumer is enjoying is a stretch. The consumer is paying more for a car they cannot legally buy without those features [14].
The size of the hedonic adjustment is contested but not trivial. BLS internal estimates suggest hedonic adjustments reduce the apparel CPI by approximately 1.9 percentage points per year and reduce the new vehicles CPI by approximately 1.0 percentage point per year [12]. Across the index as a whole the effect is smaller because the affected categories are a minority of the basket, but the directional bias is consistent: hedonic adjustment reduces reported inflation.
Geometric Weighting, 1999
In January 1999, the BLS replaced the Laspeyres (arithmetic) formula with a geometric mean formula for aggregation within most basic CPI item strata [11]. The geometric mean has a specific mathematical property: it implicitly assumes consumers maintain constant expenditure shares as relative prices change, which is mathematically equivalent to assuming an elasticity of substitution of 1 within the category.
In plain language: the geometric mean assumes consumers always substitute toward the cheaper item in a category in proportion to the price gap. Whether or not this matches actual behavior in any given category, it puts a mathematical floor on reported inflation within that category. The geometric mean of any two numbers is always less than or equal to their arithmetic mean.
The BLS’s own estimate of the impact of switching from arithmetic to geometric weighting was a reduction in measured CPI growth of approximately 0.2 percentage points per year [11]. This is on top of the substitution adjustment via Chained CPI and on top of hedonic adjustments.
Methodology Change Timeline
| Year | Change | Direction | Estimated impact (pp/year) | Source |
|---|---|---|---|---|
| 1983 | Homeowner cost: from purchase price + mortgage interest to Owners Equivalent Rent | Down | 0.5 to 1.0 in periods of asset price expansion | [4][2] |
| 1996 | Boskin Commission report delivered | (Recommendation) | (1.1 alleged overstatement) | [9] |
| 1998 | Outlet substitution adjustment | Down | ~0.1 | [9] |
| 1999 | Geometric mean aggregation within strata | Down | ~0.2 | [11] |
| 1999 onward | Expanded hedonic adjustments (electronics, vehicles, housing rent quality) | Down | ~0.2 to 0.5 (item dependent) | [12] |
| 2002 | Chained CPI-U introduced as supplemental measure | Down | ~0.25 to 0.3 vs CPI-U | [10] |
| Ongoing | Annual basket reweighting (biennial then annual) | Down on average | Small, cumulative | [11] |
Every documented major methodology change since 1980 has had a downward effect on reported inflation. There is no documented major change since 1980 that has had an upward effect. This is not a forensic claim; it is in the BLS’s own published estimates. The pattern is asymmetric.
Cumulative Effect: Alternative Measures
Several alternative inflation measures attempt to estimate what CPI would have shown had pre-1980 or pre-1990 methodology been preserved.
ShadowStats Alternate CPI (John Williams) reconstructs CPI using methodology approximately equivalent to the pre-1990 BLS approach. ShadowStats currently shows year over year inflation roughly 5 to 7 percentage points higher than reported CPI-U [15]. ShadowStats is contested by mainstream economists; the criticism is that it does not actually rebuild the index from raw price data but applies an additive adjustment to BLS data, which compounds over time. The methodology arguments ShadowStats makes are testable and largely consistent with the official BLS estimates of the impact of each methodology change. The level estimates it produces are at the high end of plausible.
Chapwood Index (Ed Butowsky) tracks the price of 500 frequently purchased items across 50 US cities. Chapwood reports five year average inflation rates of 9 to 13 percent across major US cities for the period 2014 to 2024 [16]. Chapwood explicitly does not adjust for substitution, quality, or housing methodology. It is a fixed basket of actual prices. Its methodology is a reversion to pre-1980 principles applied to a contemporary basket. It is also produced by a private firm with no peer review.
MIT Billion Prices Project (now PriceStats) tracked online prices for a representative basket and historically ran roughly in line with CPI for tradeable goods [17]. This is consistent with the position taken later in this file: CPI is reasonably accurate for tradeable goods and wildly inaccurate for housing, healthcare, and education.
Truflation (a private blockchain published index using real time retail and online price feeds) typically runs 1 to 3 percentage points above headline CPI-U during inflationary periods [18].
The point of citing these alternatives is not to assert that any one of them is “the true number.” The point is that every independent attempt to measure inflation using a fixed basket and current asset prices produces a number meaningfully higher than headline CPI-U, and the gap has widened since 1980.
Cross Check: M2 Growth, Asset Prices, Lived Purchases
The simplest sanity check on CPI is to compare it to monetary aggregates and asset prices over the same period.
| Series | January 1980 to January 2024 | Multiple | Source |
|---|---|---|---|
| M2 money stock | $1.48T to $20.84T | ~14.1x | [19] |
| Headline CPI-U | 78.0 to 308.4 | ~3.95x | [3] |
| S&P 500 (price index, dividends excluded) | ~110 to ~4900 | ~44x | [20] |
| US median existing home sale price | $63,700 to $389,000 | ~6.1x | [21] |
| Gold (USD/oz) | ~$650 to ~$2050 | ~3.2x | [22] |
| US college tuition CPI subindex | 67.4 to 904.9 | ~13.4x | [23] |
| US medical care CPI subindex | 75.4 to 562.4 | ~7.5x | [24] |
If CPI captured general price level change accurately, the gap between CPI and M2 growth would be a function of changes in money velocity and changes in real output. From 1980 to 2024, real US GDP grew approximately 3.5x [25]. If money supply grows 14x and real output grows 3.5x, a stable velocity world would predict price level growth of roughly 4x. CPI shows almost exactly that.
The problem is that velocity has not been stable, and the implicit deflator that “balances” the equation only works if you accept CPI as the price level. The asset side tells a different story: equity prices up 44x, residential real estate up 6x, gold up 3x. These are not in CPI. Owner occupied housing is in CPI as OER, not as the asset price. The 1983 OER methodology change is what makes this gap mechanically possible.
The argument made elsewhere in this knowledge base (see 04-cantillon-effect-and-wealth-transfer.md) is that money created in a fiat credit system reaches the real economy through asset markets first. CPI was reconstructed in 1983 to exclude exactly this transmission channel. The result is that “inflation” as officially measured can be low while monetary expansion is high, because the monetary expansion shows up where CPI no longer looks.
The Big Mac Index: A Tradeable Goods Cross Check
The Economist’s Big Mac Index has tracked the price of a McDonald’s Big Mac in dozens of countries since 1986 [26]. The original purpose was a purchasing power parity check on currency exchange rates. It also serves as a fixed basket inflation back check.
US Big Mac price, July 1986: $1.60 [26]. US Big Mac price, January 2024: $5.69 [26]. Multiple: ~3.6x.
Headline CPI-U over the same window (June 1986 = 109.5, January 2024 = 308.4): ~2.82x [3].
The Big Mac is a single product made from beef, bread, lettuce, sauce, labor, retail rent, and energy. It is approximately a representative basket of tradeable goods plus retail labor and rent. It has risen ~3.6x since 1986 against a CPI rise of ~2.8x. The gap is approximately 0.65 percentage points per year, which sits at the lower end of the methodology change estimates above and is consistent with CPI being roughly accurate but biased downward by 0.5 to 1.0 percentage points per year.
The Big Mac Index suggests that for a representative tradeable consumption basket, CPI is in the ballpark, with a modest persistent downward bias. It does not address the housing, healthcare, and education categories where the gap is much larger.
Sector Specific Inflation: Where the Gap Actually Lives
Headline CPI is an average. Within that average, the dispersion across categories is enormous.
| Category | 1980 to 2024 multiple | Annualized rate |
|---|---|---|
| Headline CPI-U | ~3.95x [3] | ~3.2% |
| Tradeable goods (apparel, electronics, vehicles) | ~1.5x to 2.0x [27] | ~1.0% to 1.7% |
| Food at home | ~3.5x [28] | ~3.0% |
| Energy | ~3.0x [29] | ~2.6% |
| Shelter (CPI methodology) | ~4.5x [30] | ~3.5% |
| Median home sale price (asset price) | ~6.1x [21] | ~4.2% |
| Medical care services | ~7.5x [24] | ~4.7% |
| College tuition and fees | ~13.4x [23] | ~6.1% |
| Childcare | ~5.5x [31] | ~4.0% |
Households do not consume an average. They consume housing, healthcare, education, childcare, and food. The categories that have inflated 5x to 13x since 1980 are precisely the categories that dominate the budget of working and middle class households. Tradeable goods (the categories where CPI is relatively accurate) are a smaller share of household budgets and have inflated less.
This is the “the gap is real and concentrated in non discretionary spending” claim, and it is documented in BLS data itself. It does not require a methodology critique to see. It only requires reading the subindexes.
The Lived Experience
Surveys consistently show households perceiving inflation 3 to 6 percentage points higher than headline CPI [32]. The Federal Reserve Bank of New York’s Survey of Consumer Expectations median one year inflation expectation has run roughly 1 to 4 percentage points above realized headline CPI for most of its 2013 to 2024 history [32].
The conventional academic explanation is that consumers are “biased” because they pay more attention to high salience price increases (gasoline, groceries) than to declines in apparel and electronics. The unconventional explanation is that consumers are accurately reporting the inflation rate of their actual basket, which is dominated by housing, healthcare, education, insurance, childcare, and food, all of which have inflated faster than headline CPI.
Both explanations can be partially true. The categories consumers complain about are the categories the sector specific data shows to have inflated fastest. The category bias and the salience bias point in the same direction.
International Examples
Methodology disputes are not unique to the United States. Several international cases illuminate the same dynamic.
Argentina, INDEC, 2007 to 2015. The Instituto Nacional de Estadística y Censos (INDEC) was politically captured under the Kirchner administrations and reported inflation rates of 9 to 11 percent while private and provincial measures showed 25 to 40 percent [33]. The IMF censured Argentina in 2013 over the data quality. After 2015, INDEC was rebuilt and immediately reported inflation rates consistent with the prior private measures. This is the cleanest documented case of a national statistics agency being directed to under report inflation.
Turkey, TUIK vs ENAG, 2020 onward. Turkey’s official statistics agency TUIK reports inflation rates that have diverged sharply from those reported by ENAG (Inflation Research Group), an independent academic effort [34]. In 2022, TUIK reported peak inflation around 85 percent year over year while ENAG reported approximately 185 percent. The gap is the subject of ongoing political and legal pressure on ENAG.
United Kingdom, CPI vs RPI. The UK has run two parallel inflation measures, the Retail Price Index (RPI, which includes mortgage interest payments and uses arithmetic aggregation) and the Consumer Price Index (CPI, which uses geometric aggregation and excludes housing costs differently) [35]. RPI has consistently run 0.5 to 1.0 percentage points above CPI. The UK government has progressively shifted indexed liabilities (state pensions, student loans, rail fares) from RPI to CPI on the explicit grounds that RPI is “flawed.” Critics observe that the government is the issuer of the indexed liabilities and is making the determination that the measure costing it less is the more accurate one.
Eurostat HICP. The Harmonised Index of Consumer Prices used for ECB policy excludes owner occupied housing costs entirely [36]. ECB inflation targets are set against an index that does not include the largest single household expense for most homeowners. The Eurostat OOH (Owner Occupied Housing) experimental index has been published since 2016 but has not been incorporated into the headline HICP used for policy.
The international pattern is consistent: where a government’s fiscal or monetary interest is served by lower measured inflation, methodology choices tend to deliver lower measured inflation. Where independent measures exist, they generally report higher numbers.
The Argument
The inflation tax is hidden by construction. Reported CPI permits governments and central banks to claim “low and stable inflation” while the M2 money stock has expanded fourteenfold since 1980 and asset prices have expanded by factors of three to forty depending on the asset.
Wages indexed to reported CPI lose ground continuously to two things: the asset escalator (housing, equities) which CPI excludes mechanically since 1983, and the real cost of living escalator in housing services, healthcare, and education which CPI captures only partially through OER, hedonic adjustment, and substitution adjustment.
The Boskin Commission report and the BLS methodology pages document every change openly. The methodology is not secret. The aggregate effect is also not secret: the BLS’s own estimates of the impact of each change are public. What is opaque is only that the public does not, in practice, read methodology documentation. The political economy of CPI works because the headline number is the only number the median voter, the median wage negotiator, and the median pension beneficiary ever sees.
Three concrete claims survive this analysis:
-
CPI today is not the same statistical object it was in 1980. Comparing modern CPI growth to pre-1980 CPI growth is comparing two different indices.
-
Every documented methodology change since 1980 has had a downward effect on reported inflation. The directional asymmetry is in the BLS’s own published impact estimates.
-
The cumulative gap between reported CPI and a fixed basket cost of living index for the categories households actually spend on (housing as asset, healthcare, education) is on the order of 1 to 4 percentage points per year, depending on which alternative measure and which time window is used. Over forty years this compounds to a factor of two between reported and lived inflation.
The next file in this section, 04-cantillon-effect-and-wealth-transfer.md, takes up the question of why this matters: not just that CPI under reports, but who benefits and who pays when the under reporting becomes the basis for indexing wages, pensions, and tax brackets.
Sources
[1] Bureau of Labor Statistics, Consumer Price Index home page, https://www.bls.gov/cpi/
[2] Bureau of Labor Statistics, “Changing the Treatment of Owner-Occupied Housing in the CPI,” CPI Detailed Report, January 1983; summarized in Robert Poole, Frank Ptacek, and Randal Verbrugge, “Treatment of Owner-Occupied Housing in the CPI,” BLS Working Paper, 2005, https://www.bls.gov/cpi/additional-resources/treatment-of-owner-occupied-housing.htm
[3] FRED Economic Data, Consumer Price Index for All Urban Consumers: All Items in U.S. City Average (CPIAUCSL), https://fred.stlouisfed.org/series/CPIAUCSL
[4] Bureau of Labor Statistics, “How the CPI measures price change of Owners’ equivalent rent of primary residence (OER) and Rent of primary residence (Rent),” https://www.bls.gov/cpi/factsheets/owners-equivalent-rent-and-rent.htm
[5] Bureau of Labor Statistics, “Relative Importance of Components in the Consumer Price Indexes: U.S. City Average, December 2023,” https://www.bls.gov/cpi/tables/relative-importance/2023.htm
[6] FRED Economic Data, S&P CoreLogic Case-Shiller U.S. National Home Price Index (CSUSHPINSA), https://fred.stlouisfed.org/series/CSUSHPINSA
[7] FRED Economic Data, Consumer Price Index for All Urban Consumers: Owners’ Equivalent Rent of Residences (CUUR0000SEHC), https://fred.stlouisfed.org/series/CUSR0000SEHC
[8] FRED Economic Data, Consumer Price Index for All Urban Consumers: Rent of Primary Residence (CUUR0000SEHA), https://fred.stlouisfed.org/series/CUSR0000SEHA
[9] Boskin, Michael J., et al., “Toward a More Accurate Measure of the Cost of Living,” Final Report to the Senate Finance Committee from the Advisory Commission to Study the Consumer Price Index, December 4, 1996, https://www.ssa.gov/history/reports/boskinrpt.html
[10] Bureau of Labor Statistics, “Chained Consumer Price Index for All Urban Consumers (C-CPI-U) methodology and overview,” https://www.bls.gov/cpi/additional-resources/chained-cpi.htm
[11] Bureau of Labor Statistics, “Updated Response to the Recommendations of the 1995 Advisory Commission to Study the Consumer Price Index,” BLS Report, https://www.bls.gov/cpi/additional-resources/boskin-update.pdf
[12] Bureau of Labor Statistics, “Quality Adjustment in the Consumer Price Index,” https://www.bls.gov/cpi/quality-adjustment/
[13] Gordon, Robert J., “Apparel Prices 1914 to 1993 and the Hulten/Bruegel Paradox,” NBER Working Paper No. 11548, 2005; see also Hausman, Jerry, “Sources of Bias and Solutions to Bias in the Consumer Price Index,” Journal of Economic Perspectives 17(1), 2003.
[14] Crawford, Anthony, “Measurement Bias in the Canadian CPI: An Update,” Bank of Canada Review, Spring 1998 (the regulatory mandated quality argument is developed for vehicles).
[15] Williams, John, “Alternate Inflation Charts,” ShadowStats, http://www.shadowstats.com/alternate_data/inflation-charts (cited with the standard caveat that the methodology is contested; the directional argument it makes is consistent with the BLS’s own impact estimates).
[16] Chapwood Investments, “The Chapwood Index,” https://chapwoodindex.com
[17] Cavallo, Alberto, and Roberto Rigobon, “The Billion Prices Project: Using Online Prices for Measurement and Research,” Journal of Economic Perspectives 30(2), 2016, pp. 151 to 178.
[18] Truflation, US Inflation Rate index methodology, https://truflation.com
[19] FRED Economic Data, M2 Money Stock (M2SL), https://fred.stlouisfed.org/series/M2SL
[20] FRED Economic Data, S&P 500 (SP500), https://fred.stlouisfed.org/series/SP500
[21] FRED Economic Data, Median Sales Price of Houses Sold for the United States (MSPUS), https://fred.stlouisfed.org/series/MSPUS
[22] London Bullion Market Association, Gold Price Auction historical data, https://www.lbma.org.uk/prices-and-data/precious-metal-prices
[23] FRED Economic Data, Consumer Price Index for All Urban Consumers: College Tuition and Fees (CUUR0000SEEB), https://fred.stlouisfed.org/series/CUUR0000SEEB
[24] FRED Economic Data, Consumer Price Index for All Urban Consumers: Medical Care (CPIMEDSL), https://fred.stlouisfed.org/series/CPIMEDSL
[25] Bureau of Economic Analysis, Real Gross Domestic Product, https://www.bea.gov/data/gdp/gross-domestic-product
[26] The Economist, “The Big Mac Index,” https://www.economist.com/big-mac-index ; historical data also archived at https://github.com/TheEconomist/big-mac-data
[27] Bureau of Labor Statistics, Consumer Price Index detailed report tables, Apparel and Transportation Commodities subindexes, https://www.bls.gov/cpi/tables/
[28] FRED Economic Data, Consumer Price Index for All Urban Consumers: Food at Home (CUUR0000SAF11), https://fred.stlouisfed.org/series/CUUR0000SAF11
[29] FRED Economic Data, Consumer Price Index for All Urban Consumers: Energy (CPIENGSL), https://fred.stlouisfed.org/series/CPIENGSL
[30] FRED Economic Data, Consumer Price Index for All Urban Consumers: Shelter (CUSR0000SAH1), https://fred.stlouisfed.org/series/CUSR0000SAH1
[31] Bureau of Labor Statistics, Consumer Price Index for All Urban Consumers: Day Care and Preschool (CUUR0000SEEB03), https://www.bls.gov/cpi/
[32] Federal Reserve Bank of New York, Survey of Consumer Expectations, https://www.newyorkfed.org/microeconomics/sce
[33] International Monetary Fund, “Statement by the IMF Executive Board on Argentina,” Press Release No. 13/33, February 1, 2013, https://www.imf.org/en/News/Articles/2015/09/14/01/49/pr1333
[34] ENAG (Inflation Research Group of Turkey), monthly inflation reports, https://enag.org.tr/
[35] UK Office for National Statistics, “Shortcomings of the Retail Prices Index as a measure of inflation,” https://www.ons.gov.uk/economy/inflationandpriceindices/articles/shortcomingsoftheretailpricesindexasameasureofinflation/2018-03-08
[36] Eurostat, “Harmonised Index of Consumer Prices (HICP) Methodological Manual,” https://ec.europa.eu/eurostat/web/hicp/methodology
[37] Bureau of Labor Statistics, Argentina national statistics office (INDEC), https://www.indec.gob.ar