Listen to the Audio Mises Wire version of this article.
Modern macroeconomics has made price stability the primary objective of monetary policy. It is assumed that central banks can ensure price stability by skillfully managing the money supply, thereby creating the conditions for economic growth and prosperity.
In order to provide a safety buffer against the dreaded price deflation, central banks around the world try to generate a positive but moderate rate of price inflation. Price stability thus means a stable rate of inflation. The prices of goods and services should on average rise slowly at a constant rate over the medium and long term. In the eurozone, the aim is to achieve a price inflation rate close to but below 2 percent.
However, it cannot be denied that measuring a general price level and its rate of change is associated with major problems. The formal inflation target of the central banks must be operationalized in practice. It is therefore necessary to determine which prices are targeted and how they are to be summarized in a weighted average.
The Harmonized Index of Consumer Prices
The member states of the eurozone have agreed on a standardized procedure for measuring inflation. The Harmonized Index of Consumer Prices (HICP) is the operationalized target variable of monetary policy. The calculation of the HICP is relatively complex, as attempts are made to eliminate possible distortions in the measurement of inflation by means of elaborate procedures and estimates. However, it is highly questionable whether this is successful. In the following, I would like to take a closer look at two important sources of bias.
The HICP consists of twelve subindices which group together different classes of goods. Each of the subindices consists of different subcategories, which are again subdivided until the individual prices of certain goods and services are reached at the lowest level. These unit prices must be adequately weighted for the calculation of the index. The principle is that goods and services on which a large proportion of income is spent must be given a higher weighting than those goods and services that are purchased only very sporadically and in small quantities. Formally, therefore, the weights are determined by the real turnover shares. In Germany, for example, the weight of the subindex “Food and non-alcoholic beverages” (CP01) currently stands at 11.3 percent, which means that the average German household is assumed to spend 11.3 percent of its consumption expenditure on goods in this category. By comparison, the weight for “Alcoholic beverages, tobacco and narcotics” (CP02) is 4.2 percent.
As consumption decisions are constantly changing, the weighting scheme applied may lead to distortions in the measurement of inflation. In the 1990s, for example, the Boskin Commission found a systematic overestimation of price inflation rates in the US of 0.4 percentage points per year. The cause of the distortion was the systematic substitution behavior of households.
The argument is as follows. Let us assume a base year with a given weighting scheme for all individual goods and services included in the index. This weighting scheme reflects the consumption behavior of households in the base year. This behavior changes over time, partly because prices for some goods rise faster than for others. Over time, households will tend to buy less of those goods whose prices rise faster. And they will instead buy more of other goods that have remained relatively cheap. Households will thus substitute goods with a relatively high rate of inflation for goods with a relatively low rate of inflation. If the weighting scheme is not changed, an upward distortion of the measured price inflation will result. One would overestimate price inflation.
Let me illustrate this with a simple example. Imagine a price index for soft drinks. The purchase prices of Coke and Pepsi are included in the index at 50 percent each, because on average households spend proportionally the same amount on both beverages. Assume that over a certain period of time the price of Pepsi increases by 5 percent per year. The price of Coke increases by only 1 percent annually. If the weighting is not changed, the overall inflation rate is 3 percent. In fact, however, the consumption behavior has shifted due to the different inflation rates. On average, households now buy more Coke and less Pepsi. Let us assume that households now spend four times as much on Coke as on Pepsi. The weighting would therefore have to be adjusted so that Coca Cola is included in the index at 80 percent and Pepsi at only 20 percent. The adjusted annual inflation rate, using the new weighting scheme, would therefore be 1.8 percent instead of 3 percent.
As a result of this reasoning, the weighting scheme of the HICP is now continuously adjusted, with the result that reported price inflation is lower than it would have been otherwise. Let us disregard any possible inaccuracies and assume that the adjustments to the weighting scheme perfectly reflect changing consumer behavior. Wouldn’t this be overlooking a crucial point?
The answer is yes. If consumers do not switch to other products because of changing preferences, but simply because the prices of the products they would actually prefer have risen disproportionately, then consumers are worse off. The economist would call this a welfare loss. This welfare loss corresponds to a real increase in the cost of living, which is not reflected in the official figures if the weighting scheme in the index is adjusted accurately according to the changing consumption decisions of households. We end up with a downward distortion of the measured price inflation. The rate of inflation is then underestimated.
The second major sources of distortions in official inflation statistics are changes in the quality of goods. Here, too, the Boskin Commission of the 1990s found an upward bias of 0.4 percentage points per year in the US because quality improvements in products were not adequately priced in. The measured inflation rate was therefore once again too high.
The theoretical argument is compelling. Assume that prices do not change over a given period of time, but that the quality of the goods increases steadily. Then consumers get better quality for the same money. If you now say that the inflation rate is 0 percent, you are exaggerating. In fact, ceteris paribus the standard of living has improved: you get more quality for the same money, or the same quality for less money. Hence, the reported inflation rate should be negative.
In the following period, not only in America, but also in Europe, so-called hedonic methods of quality adjustment were introduced. For many products, therefore, not only are the observed purchase prices included in the index, but adjusted prices that are supposed to reflect the quality changes. In the following, I would like to address only two fundamental problems with quality adjustment.
First, producers have an incentive to highlight the quality improvements in the products they sell. When a car or computer becomes more powerful or faster, this can be seen in measurable core values. The car has more horsepower. The computer has a faster CPU. Manufacturers will openly communicate these core values and use them to promote their products. Quality improvements will thus be made transparent and comprehensible for buyers. They can therefore also be taken into account relatively easily in official statistics.
On the other hand, producers have an incentive to conceal possible deteriorations in quality from buyers. If the casing and wiring of a computer are made of inferior material, this is usually not mentioned in the product description. If you want to detect deteriorations in quality, you often have to look very closely. In many cases, they are not easily detectable and cannot be quantified.
This leads to a systematic distortion. On the one hand, quality improvements are visible and taken into account. The prices of the products in question are reduced in the official statistics. Quality deteriorations, on the other hand, remain undetected and the prices of the products concerned are not increased accordingly. It is therefore probable that the adjustments made here also create a downward bias. The official statistics then report a price inflation rate that is too low.
The second point I would like to add has not yet been taken into account at all in the relevant literature. Let us assume that all quality changes are accurately priced in by official statistics. Even if this were the case, it would create a downward bias in reported price inflation. The reason for this is that a given quality improvement in a product already creates deflationary price pressures on other goods without any adjustment being made at all. This pressure arises in particular for the previously common and now inferior predecessors of the new product.
For example, when Apple launched the very first smartphone on the market, the iPhone, a negative price pressure on conventional mobile phones arose, because Apple dug away market shares from competing mobile phone manufacturers with its new product. As a result, competitors were forced to charge lower prices for their products than would otherwise have been the case. Only by offering lower prices could at least some buyers be convinced not to switch to the new iPhone.
This negative price pressure on competing products, which results from innovation, is already reducing the measured inflation rates. This means that a given improvement in quality is partly reflected in falling prices for other goods. If the price of the quality-improved good is adjusted in addition to this market adjustment (assuming that it is even possible to do this accurately), we would overshoot the mark. Price inflation would be underestimated.
There is no doubt that both substitution effects and changes in the quality of goods and services pose practically insoluble problems for official inflation statistics. Quality changes cannot be quantified objectively. This circumstance alone opens up enormous discretionary scope for official price statistics, which also has an impact on monetary policy. The M1 money supply in the euro area has increased more than fivefold since its inception. This could also be politically justified, because the reported price inflation was relatively low. Prices in the euro area have officially increased by only slightly more than 40 percent since 1999. Is price inflation systematically underestimated? The suspicion is obvious.
Even if the practical problems of measuring inflation, which arise from substitution effects and quality changes, could be solved sufficiently well, the application of the procedures currently used in official statistics would lead to a systematic underestimation of price inflation. The upward biases, which are undoubtedly relevant, once identified are reversed into downward biases when we consider other factors. On both points—substitution effects and quality changes—the results would overshoot the mark, even if the current methods could be applied accurately and flawlessly.
In addition, there are other gaps in the official measurement of inflation. Asset prices are not taken into account. However, disproportionate price inflation has been taking place in recent decades, especially for long-term assets such as real estate and stocks. It is not surprising that the median of subjectively perceived price inflation rates in the eurozone is 5 percentage points higher per year than the officially reported inflation rate.