CPI Biases from Substitution and New Goods

Why a Price Index Can Differ from the Cost of Living

A consumer price index tracks price changes using a specified basket of goods and services. That basket makes comparisons possible: the index asks how much the same collection would cost at different dates. But households do not necessarily keep buying the same quantities when relative prices change. They may substitute toward goods that have become relatively cheaper, and they may gain access to products that were not available when the basket was set.

These changes matter because the cost of living is not simply the cost of buying an unchanged list of items. In economic terms, it is the minimum spending required to reach a specified level of well-being, given prices and available choices. A fixed-basket CPI and a cost-of-living index therefore answer related but different questions. The fixed basket is transparent and useful for measuring price change, but it can miss some ways consumers adjust.

Definition: A fixed-basket price index compares the cost of a specified basket at current prices with its cost at base-period prices. A cost-of-living index compares the minimum expenditure needed at different prices to attain the same reference level of utility, or well-being. The two measures can differ because consumers can change what they buy and, over time, which goods are available.

The standard fixed-basket index using base-period quantities is the Laspeyres price index. Let $p_{j0}$ and $p_{jt}$ be the prices of good $j$ in the base period and period $t$, and let $q_{j0}$ be its base-period quantity. The index is:

P_L = \frac{\sum_j p_{jt}q_{j0}}{\sum_j p_{j0}q_{j0}}.

An index value of $1.12$, for example, means the fixed basket costs 12% more than it did in the base period. The index does not say that every household’s spending rose by 12%, or that each household bought the same quantities. It reports the price change for the specified basket.

Substitution and the Fixed-Basket Effect

Suppose two goods serve related purposes. If the price of one rises relative to the other, a consumer may buy less of the now-relatively-expensive good and more of its substitute. A fixed-basket calculation continues to price the original quantities, even though a consumer may change those quantities and spend less than the fixed basket would cost while still reaching the same level of well-being.

This leads to the usual substitution-bias result: under standard consumer-choice assumptions, a base-period fixed-basket index is at least as large as the cost-of-living index for maintaining base-period utility. This is a comparison with a particular theoretical measure, not a claim that every published CPI overstates every household’s experienced inflation. The size and direction of the difference depend on the basket, preferences, available alternatives, and the index’s construction.

Fix the reference utility.
Let the consumer’s base-period bundle provide utility $U_0$. Assume the consumer chooses a least-cost bundle at each set of prices and preferences do not change.

Compare costs at current prices.
The old bundle is still one possible way to attain $U_0$. Its current cost is $\sum_j p_{jt}q_{j0}$. A consumer who can choose quantities may find a different bundle that also reaches $U_0$ for no more than that amount.

Divide by base-period cost.
If the base bundle was a least-cost choice, its cost is the minimum base-period expenditure for $U_0$, namely $\sum_j p_{j0}q_{j0}$. Dividing the current minimum cost by this base cost gives a cost-of-living index no greater than $P_L$.

The key inequality follows because the base bundle is feasible at current prices, while the consumer’s minimum-cost choice can be any bundle that reaches the target utility:

\min_{\{q:\,U(q)\ge U_0\}}\sum_j p_{jt}q_j \ \le\ \sum_j p_{jt}q_{j0}.

If base-period spending also minimized the cost of reaching $U_0$, then dividing both sides by that base-period minimum expenditure gives the cost-of-living index on the left and the Laspeyres index on the right. The result relies on the assumptions just stated. It is not a universal numerical correction that can be applied to a CPI without information about preferences and feasible choices.

Worked Example: Substitution Lowers the Cost of Maintaining Utility

A hypothetical consumer buys two goods, $x$ and $y$. Preferences are represented by $U(x,y)=\sqrt{xy}$. In the base period, prices are $p_{x0}=\$2$ and $p_{y0}=\$4$, and the consumer chooses $x=2,\ y=1$. In the next period, both prices are $\$3$. Compare the fixed-basket inflation rate with the cost change for maintaining base-period utility.

Method. First calculate the Laspeyres index using the base bundle. Then find the minimum cost of attaining the same utility at current prices. The calculation assumes the stated utility function remains unchanged and both goods can be purchased in the stated quantities.

The base bundle gives utility $U_0=\sqrt{2\cdot1}=\sqrt{2}$. Its base-period cost is $(\$2)(2)+(\$4)(1)=\$4+\$4=\$8$. At current prices, the same fixed bundle costs $(\$3)(2)+(\$3)(1)=\$6+\$3=\$9$. Thus:

P_L=\frac{9}{8}=1.125, \qquad \text{fixed-basket inflation}=1.125-1=0.125=12.5\%.

To find the least cost of reaching $U_0$, impose $xy=U_0^2=2$. With both prices equal to $\$3$, the least-cost bundle has equal quantities: $x=y$. Since $xy=2$, this gives $x=y=\sqrt{2}$. Its cost is $(\$3)\sqrt{2}+(\$3)\sqrt{2}=6\sqrt{2}$, or approximately $\$8.49$. The base-period minimum cost was $\$8$, so the cost-of-living index is:

\frac{6\sqrt{2}}{8}\approx1.0607, \qquad \text{cost-of-living increase}\approx 6.1\%.

The consumer shifts from the base bundle $(2,1)$ toward equal quantities when the prices become equal. The fixed basket rises by 12.5%, while the minimum expenditure to maintain utility rises by about 6.1%. The approximate dollar amount is rounded to the nearest cent, and the rates to one decimal place. Common mistake: using the cost of the old bundle as though the consumer cannot adjust quantities. That gives the Laspeyres result, not the minimum cost of maintaining utility.

New Goods and the Value of Choice

A new good can affect the cost of living even before it appears in a regularly updated basket. If consumers gain access to an additional option, they can choose it when it offers better value than the existing alternatives. A price index that continues to represent only the older choices may not capture the benefit of having that additional option.

The timing of inclusion matters. Before a new product is available, a household cannot use it to meet its needs. Once it becomes available, the household can choose between it and existing goods. If the new option provides the same relevant service at lower cost, the minimum spending needed to reach a given level of well-being can fall, even if the prices of the older goods have not changed. A basket that does not yet include the new option can miss this change in opportunity.

Worked Example: A New Alternative Is Not Yet in the Basket

In a hypothetical market, the only available good in the base period is $A$, which costs $\$10$ per unit. One unit supplies one unit of a particular service. In the next period, a new good $B$ becomes available. It supplies the same service, and costs $\$8$ per unit. Good $A$ still costs $\$10$. For simplicity, assume one unit of either good provides exactly the same utility, and the consumer can buy either in whole units. Compare the old-good fixed-basket result with the minimum cost of obtaining one unit of service.

Method. Price the base-period basket using its original good, then compare the minimum costs of one unit of service across periods. This isolates the effect of the expanded choice set; it does not assume that every new product is cheaper or equally useful.

The base basket contains one unit of $A$, so its base cost is $\$10$. If the index continues to price that basket, its current cost is also $\$10$. The fixed-basket index is therefore $\$10/\$10=1$, implying 0% inflation for that basket.

But in the next period, the consumer can obtain one unit of service with one unit of $B$ for $\$8$, rather than spending $\$10$ on $A$. The minimum-cost ratio for this target is $\$8/\$10=0.8$, a 20% decrease. In this hypothetical case, the older-good basket records no price change while the lowest cost of obtaining the specified service falls by 20%. The calculations are exact. Common mistake: concluding that the cost of living cannot change because the price of the old good is unchanged. The available options changed, so the least-cost choice changed too.

This example deliberately uses a simple, exact comparison. In actual measurement, a new product may differ in features, usefulness, or the consumers who purchase it. The benefit of an expanded choice set cannot automatically be measured by subtracting the new product’s sticker price from an old price. The relevant question is how the newly available option affects the cost of attaining a comparable level of well-being. Adjusting for differences in product quality is a separate measurement problem.

Changing Choices and Different Basket Results

A current-basket index uses current quantities rather than base-period quantities. In its standard Paasche form, it compares the current cost of the current basket with what that same basket would have cost at base-period prices. Because it uses different quantities, it can produce a different result from the Laspeyres index. This difference helps illustrate the role of changing consumer choices; it does not, by itself, establish which index equals the cost of living.

P_P=\frac{\sum_j p_{jt}q_{jt}}{\sum_j p_{j0}q_{jt}}.

The following example uses only two goods. The arithmetic shows how the quantity weights affect each index; it does not assume that the two baskets deliver exactly the same utility.

Worked Example: Base and Current Baskets Give Different Indexes

A hypothetical household buys goods $A$ and $B$. In the base period, $A$ costs $\$5$, $B$ costs $\$10$, and the household buys 4 units of $A$ and 2 of $B$. In the current period, $A$ costs $\$8$, $B$ costs $\$8$, and the household buys 2 units of $A$ and 4 of $B$. Calculate both price indexes.

Method. The Laspeyres index prices the base quantities in both periods. The Paasche index prices the current quantities at both sets of prices.

The base basket costs $(\$5)(4)+(\$10)(2)=\$20+\$20=\$40$ in the base period. At current prices, it costs $(\$8)(4)+(\$8)(2)=\$32+\$16=\$48$. Thus the Laspeyres index is $\$48/\$40=1.20$, or 20% inflation.

The current basket costs $(\$8)(2)+(\$8)(4)=\$16+\$32=\$48$ at current prices. At base prices, those same current quantities would have cost $(\$5)(2)+(\$10)(4)=\$10+\$40=\$50$. Thus the Paasche index is $\$48/\$50=0.96$, or a 4% decrease for that current basket. Both calculations are exact.

The large difference comes from the baskets’ different weights: $A$ became relatively more expensive, while $B$ became relatively cheaper, and the current basket contains more $B$. Common mistake: dividing the base basket’s current cost by the current basket’s base cost and treating the result as one standard index. Each index must use the same basket in its numerator and denominator.

Interpreting CPI Bias Carefully

“Bias” here means a systematic difference between a measured index and a specified target, such as the cost of maintaining base-period utility. It does not mean that a CPI calculation is arithmetically wrong. A fixed basket can be the right tool for answering the question, “How did the cost of this basket change?” The concern arises when the result is interpreted as an exact measure of the change in every household’s cost of maintaining its standard of living.

Substitution and new-goods effects should be treated as potential sources of divergence, not as automatic corrections. Substitution can be limited if alternatives are poor, if quantities cannot adjust, or if the household’s needs require a particular item. A new product may not reduce the cost of obtaining the same service. Households also differ in what they buy and which new options they can use. These points help explain why a single index can be informative while still not matching every household’s experience.

Substitution: a fixed base basket continues to assign weight to goods at their old quantities, even when consumers change their choices in response to relative-price changes.
New goods: delayed inclusion can leave out benefits from an expanded set of options, especially when a new option provides a comparable service at lower cost.
Different questions: the Laspeyres index prices the base basket; a cost-of-living index targets the expenditure needed to maintain utility. They should not be treated as interchangeable without assumptions.

For interpretation, first identify the basket and period weights behind an index. Then ask whether the intended conclusion concerns the cost of that basket or the spending needed to maintain a particular level of well-being. Keeping those questions separate prevents a common misreading: treating measured inflation as an exact, universal change in household living costs.

Key takeaway: A fixed-basket CPI measures price change for a specified basket. When consumers substitute toward relatively cheaper goods or gain access to new choices, the cost of maintaining a given level of well-being may change differently. The direction and size of the difference depend on the assumptions and choices being compared.

Check Your Understanding

For each problem, identify the relevant basket or choice set before calculating.

A base basket contains 3 units of $A$ at $\$4$ each and 2 units of $B$ at $\$6$ each. Current prices are $\$5$ for $A$ and $\$5$ for $B$. Find the Laspeyres inflation rate. (Answer: 4.17%)
A household needs one unit of service. In the base period, the only available product costs $\$12$ per unit. In the current period, that product still costs $\$12$, but a new, equally useful product is available for $\$9$. Compare the old-product fixed-basket inflation rate with the minimum-cost change for the service. (Answer: Fixed-basket inflation is 0%; minimum cost falls 25%, from $\$12$ to $\$9$.)
A base basket contains 3 units of $A$ at $\$2$ each and 2 units of $B$ at $\$5$ each. Current prices are $\$3$ for $A$ and $\$4$ for $B$. Calculate the Laspeyres index and inflation rate. (Answer: Index $17/16=1.0625$; inflation is 6.25%.)
In one sentence, state why the Laspeyres index is at least as large as the cost-of-living index under the assumptions used in the derivation. (Answer: The old basket can attain the reference utility, while the least-cost choice among all bundles attaining that utility cannot cost more than that feasible basket.)
Base prices for $A$ and $B$ are $\$4$ and $\$2$, and base quantities are 2 and 3. Current prices are $\$5$ and $\$1$, and current quantities are 1 and 5. Calculate the Laspeyres and Paasche indexes. (Answer: Laspeyres $13/14\approx0.9286$, or $-7.14\%$; Paasche $10/14\approx0.7143$, or $-28.57\%$.)

What You'll Learn