Introduction: The Distinction That Makes GDP Usable
If Zimbabwe’s nominal GDP rose by 10,000% in a year, nobody would conclude that Zimbabweans were 100 times richer. Yet that is precisely what an uncorrected GDP figure would imply.
Nominal GDP mixes two things that must be separated: how much stuff was produced, and how much that stuff cost. Real GDP isolates the first. The GDP deflator isolates the second. Everything in macroeconomics that concerns growth, recession or living standards depends on getting this decomposition right.
1. Nominal GDP
Nominal GDP — the value of final output measured at current prices, i.e. the prices prevailing in the year the output was produced. Also called GDP at current prices, or money GDP.
Nominal GDP changes for two entirely different reasons — a change in Q, or a change in P — and it cannot tell you which. It is therefore useless for comparing living standards across time.
It is not useless in general. Nominal GDP is the correct denominator for any ratio whose numerator is also nominal: debt-to-GDP, tax-to-GDP, deficit-to-GDP. Deflating one and not the other is a classic error.
2. Real GDP
Real GDP — the value of final output measured at constant prices from a chosen base year. It holds prices fixed so that any change in the index reflects a change in the physical quantity of output.
| Year | Price of Bread | Bread Qty | Price of Cars | Car Qty |
|---|---|---|---|---|
| 2020 (base) | $2 | 100 | $10,000 | 5 |
| 2025 | $3 | 120 | $15,000 | 6 |
Nominal GDP 2020 = (2 × 100) + (10,000 × 5) = $50,200
Nominal GDP 2025 = (3 × 120) + (15,000 × 6) = $90,360 → nominal growth of 80.0%
Real GDP 2025 (at 2020 prices) = (2 × 120) + (10,000 × 6) = $60,240 → real growth of 20.0%
GDP Deflator 2025 = (90,360 ÷ 60,240) × 100 = 150.0 → prices rose 50% since 2020
Notice the arithmetic: 1.20 × 1.50 = 1.80. Real growth × price growth = nominal growth, exactly.
3. The GDP Deflator
Rearranged, this gives the formula you will use more often:
The deflator uses current-period quantities as weights, because those are the quantities actually produced this year. The CPI uses base-period quantities, held fixed.
This is not a technicality. A Laspeyres index (CPI) tends to overstate inflation, because it ignores consumers substituting away from goods that have become expensive. A Paasche index (deflator) tends to understate it, because it weights by the post-substitution basket. Modern statistical agencies increasingly use chained indices — updating weights each period and linking — which sit between the two, and are why real GDP series are frequently revised.
4. GDP Deflator versus CPI
| GDP Deflator | CPI | |
|---|---|---|
| Index type | Paasche (current weights) | Laspeyres (fixed weights) |
| Coverage | All domestically produced final goods | Goods and services bought by households |
| Imports | Excluded | Included |
| Capital goods | Included | Excluded |
| Government services | Included | Largely excluded |
| Basket updated | Automatically, every period | Periodically, by survey |
| Frequency | Quarterly | Monthly |
Q: A country imports all its oil. World oil prices double. What happens to its CPI and its GDP deflator?
A: CPI rises sharply — households pay more at the pump, and imported petrol is in the consumption basket. The GDP deflator barely moves — oil is not domestically produced and is therefore outside the deflator’s scope. The two measures can diverge substantially, and did so across 2022. Examiners love this question because it tests whether you understand what each index is for, rather than just its formula.
5. Base Year Choice and the Substitution Problem
Real GDP growth depends on the base year chosen. Different base years give different growth rates for the same underlying economy.
Why? Because goods whose prices fall fastest (computers, semiconductors) tend to be the goods whose quantities rise fastest. A distant base year assigns those goods their old, high prices — inflating their contribution to real growth. A recent base year assigns them low prices — deflating it. This is substitution bias operating on the output side, and it is why the US moved to chain-weighted real GDP in 1996.
The problem: Using a 1987 base year, US real GDP in the mid-1990s valued each computer at its 1987 price — enormous. Computer output was exploding. The result was an implausibly large computer contribution to measured real growth.
The reform: In 1996 the Bureau of Economic Analysis adopted chain-weighted real GDP, using a Fisher ideal index — the geometric mean of Laspeyres and Paasche — with weights updated annually and linked.
The consequence: Measured growth in the late 1980s and 1990s was revised. Chain-weighted real GDP components no longer add up to the total (they must be presented as “contributions to growth” instead), which is confusing but unavoidable.
The deeper point: There is no theory-free way to aggregate heterogeneous physical quantities into a single number called “output.” Every method embeds an index-number assumption. Real GDP is a construct, and reasonable statisticians using defensible methods will disagree about it.
Griliches, Z. (1961), “Hedonic Price Indexes for Automobiles”; Nordhaus, W. (1996), “Do Real-Output and Real-Wage Measures Capture Reality?”
The problem: A 2025 smartphone at the same price as a 2015 smartphone is not the same good. If we record “no price change,” we have implicitly recorded “no quality improvement” — and understated real output growth.
The method: Hedonic regression treats a good as a bundle of characteristics and regresses price on those characteristics (processor speed, memory, screen size). The coefficients yield an implicit price per unit of quality, allowing statisticians to separate genuine price change from quality change.
Nordhaus’s illustration: He traced the true price of light — lumens per hour — from open fires through candles, gas lamps, and filament bulbs to fluorescents. Measured by conventional price indices, lighting became modestly cheaper over two centuries. Measured in cost per unit of actual illumination, it became cheaper by orders of magnitude. Conventional indices had missed almost the entire gain.
Why it matters: If Nordhaus is right that price indices systematically miss quality improvement, then historical real growth has been understated and historical inflation overstated. The revision would be enormous — and it is not applied uniformly across sectors, which means measured productivity differences between, say, computing and healthcare may partly reflect differences in how hard quality is to measure.
6. Exam Technique
- Compute nominal GDP, real GDP, and the deflator from a table of prices and quantities. This appears almost every year.
- In the base year, nominal GDP = real GDP, and the deflator = 100. Always. Use this as a check on your arithmetic.
- Convert between nominal and real: Real = (Nominal ÷ Deflator) × 100.
- Real growth rate ≈ nominal growth rate − inflation rate. (Exact only for small rates; the exact relation is multiplicative.)
- When asked to compare living standards across time or countries, insist on real GDP per capita, adjusted for purchasing power parity. Each of those three adjustments corrects a distinct error.
- Evaluate the deflator-versus-CPI distinction using the imported-oil example. It demonstrates conceptual command, not memorisation.
- Note that real GDP is base-year dependent, and that chain weighting exists precisely because that dependence is a defect.
- Forgetting that in the base year the deflator is 100 by construction.
- Using base-year quantities when computing real GDP. You use base-year prices and current-year quantities.
- Deflating a nominal ratio like debt-to-GDP. Both numerator and denominator are nominal; the ratio is already real.
- Treating “real GDP growth = nominal growth − inflation” as exact. It is a first-order approximation.
- Assuming CPI and the deflator should give the same inflation rate. They measure different baskets and routinely diverge.
Summary
Nominal GDP measures output at today’s prices and conflates quantity with price. Real GDP holds prices fixed and isolates quantity. The GDP deflator is the ratio between them, and is therefore a price index derived from the national accounts rather than from a household survey.
The decomposition is essential and it is imperfect. Real GDP depends on a base year; on how quality change is handled; on whether the index is Laspeyres, Paasche or chained. These are not marginal quibbles — they materially alter measured growth. The number that governments live and die by is, at its foundation, an index-number problem with no unique solution.
Exercise 1 — Is There a Fact of the Matter About Real Growth?
A Laspeyres index overstates inflation; a Paasche index understates it; the Fisher ideal index is the geometric mean of the two and has no clear behavioural interpretation. Chain-weighted real GDP components no longer sum to the total.
Given that different defensible index-number choices produce different real growth rates for the same economy, is “real GDP growth” a fact about the world, or an artefact of measurement convention? If the latter, what justifies using it to declare recessions, index pensions, and evaluate governments? Consider whether the existence of measurement ambiguity is fatal, or whether it is simply the ordinary condition of all empirical science.
📄 Read: Diewert, W. E. (1976). “Exact and Superlative Index Numbers.” Journal of Econometrics, 4(2), 115–145. Diewert’s concept of a “superlative” index — one exact for a flexible aggregator function — is the theoretical justification for chain weighting, and it repays close reading.
Exercise 2 — The Price of Light
Nordhaus showed that conventional price indices, tracking the price of light bulbs rather than the price of light, missed nearly the entire two-century decline in the cost of illumination. If this error is representative, historical real growth has been substantially understated.
But hedonic adjustment is applied unevenly: heavily to computers, barely to healthcare or education. Does this mean that measured productivity growth in computing is overstated relative to healthcare — or that healthcare’s real quality gains are simply invisible? Design a hedonic specification for a hospital’s output and identify precisely where it fails.
📄 Read: Nordhaus, W. D. (1996). “Do Real-Output and Real-Wage Measures Capture Reality? The History of Lighting Suggests Not.” In Bresnahan & Gordon (eds.), The Economics of New Goods. University of Chicago Press, 27–70.
References
- Diewert, W. E. (1976). Exact and Superlative Index Numbers. Journal of Econometrics, 4(2), 115–145.
- Griliches, Z. (1961). Hedonic Price Indexes for Automobiles: An Econometric Analysis of Quality Change. In The Price Statistics of the Federal Government. NBER.
- Landefeld, J. S., Seskin, E. P., & Fraumeni, B. M. (2008). Taking the Pulse of the Economy: Measuring GDP. Journal of Economic Perspectives, 22(2), 193–216.
- Nordhaus, W. D. (1996). Do Real-Output and Real-Wage Measures Capture Reality? The History of Lighting Suggests Not. In T. F. Bresnahan & R. J. Gordon (eds.), The Economics of New Goods. University of Chicago Press, 27–70.
- United Nations, European Commission, IMF, OECD & World Bank (2009). System of National Accounts 2008.
- Whelan, K. (2002). A Guide to U.S. Chain Aggregated NIPA Data. Review of Income and Wealth, 48(2), 217–233.