Price-setting Microdata Analysis Network (PRISMA)

How much does quality adjustment matter in measuring consumer price inflation? To address this question, we use different sources of micro and macro price data for Germany and the euro area. For Germany, we find that quality adjustment applies to a large range of goods and services but, on average, price adjustments due to quality changes reduce headline inflation only by 0.06 percentage points, which is balanced out by an increase due to quantity adjustment (e.g. a smaller package size) of the same amount. For the euro area, we assess the impact of heterogeneous quality adjustment methods by deriving the distribution of member states’ cumulative inflation rates for typical quality-adjusted products. Our macro-based estimate makes up to ± 0.2 percentage points for headline HICP inflation and ranges between± 0.1 and 0.3 percentage points for core inflation, when controlling for income differentials between member states. Finally, we illustrate the role of heterogeneous quality adjustment methods in the euro area based on micro price data for washing machines. We show that the price development of this product would have been lower by about 3.5 percentage points during the first years of the euro area and by about half a percentage point during recent years, if prices in the member states had been quality-adjusted in exactly the same way.

Using web-scraping, we have gathered daily data on prices and product availability for over 50,000 unique food items from 7 major retailers in Poland since January 2020. We find a 24% drop in product availability during the first COVID-19 wave in April 2020. Next, with two-step and iterative GMM applied to a variety of specifications, we show a modest price response to stockouts, as a 10% decline in food product availability leads to an immediate 0.5% rise in prices throughout the first COVID-19 wave. During the subsequent pandemic waves, this link disappears. We also detect only a 4% drop in product availability following the onset of the Russian invasion of Ukraine in February 2022, thus we suspect that the recent price movements are linked primarily to cost and demand factors. Our evidence is indicative of retailers fearing consumer anger during unorthodox times.

The consensus in the literature on providing accurate inflation forecasts underlines the importance of precise nowcasts. In this paper, we focus on this issue by employing a unique, extensive dataset of online food and non-alcoholic beverages prices gathered automatically from the webpages of major online retailers in Poland since 2009. We perform a real-time nowcasting experiment by using a highly disaggregated framework among popular, simple univariate approaches. We demonstrate that pure estimates of online price changes are already effective in nowcasting food inflation, but accounting for online food prices in a simple, recursively optimized model delivers further gains in the nowcast accuracy. Our framework outperforms various other approaches, including judgmental methods, traditional benchmarks, and model combinations. After the outbreak of the COVID-19 pandemic, its nowcasting quality has improved compared to other approaches and remained comparable with judgmental nowcasts. We also show that nowcast accuracy increases with the volume of online data, but their quality and relevance are essential for providing accurate in-sample fit and out-of-sample nowcasts. We conclude that online prices can markedly aid the decision-making process at central banks.

It has been widely documented that, depending on their consumption, households experience different inflation rates which are concealed in aggregate price indices such as the CPI. Using micro price data from a household scanner data panel for Austria spanning from 2008 to 2018, we analyse price dynamics faced by households and try to explain some of the causes for inflation differences across households. By identifying not only the products consumed but also the specific prices paid by the households we are able to go beyond what is done in most of the previous literature on inflation heterogeneity. We find a considerable and persistent degree of heterogeneity in household inflation rates with an interquartile range of about 6 percentage points which is way more than what has been found in previous studies for Austria. Considering average prices at the product level instead, the degree of inflation heterogeneity drops by two thirds which suggests that the prices paid by households are an important source of inflation heterogeneity besides different consumption baskets. We also find that inflation tends to be higher for households with older household members and those living in the West of Austria, while shopping intensity and higher household income appear to have a dampening impact on household inflation rates.

(Why) do prices and inflation rates differ within the euro area? We study the relevance of a national border for grocery prices in the otherwise homogenous and highly integrated border region of Austria and Germany. Using transaction data on prices and quantities from a large household panel we compare the prices of identical products within a narrow band along the Austrian-German border. We find large assortment and price differences between these two regions. Even within multinational retail chains the prices of identical products on the two sides of the border differ on average by about 23%. However, these price differences are not very persistent over time indicating little arbitrage gain by undifferentiated cross-border shopping. Ensuing product-level inflation rates differ only for half of the chains between the two countries. The results highlight the importance of the history-dependent evolution of distribution networks and of the structure of sales organisation as a driver of price and inflation heterogeneity.

In this paper we complement the existing literature on price rigidity using webscraped data from online shops in Austria for the period from January 2021 to date. We calculate the major price adjustment statistics (frequency and size of price changes) with thousands of online prices observed at daily frequency. To compare the period before the inflation surge with the ensuing high inflation period, we split the sample into two parts: January to August 2021 and September 2021 to August 2022. As comprehensive and reliable daily price data are available in our dataset only for online supermarkets, we limit our analysis mainly to grocery items. The resulting statistics on the frequency and size of price changes are measured at daily basis and have to be interpreted as such. Our preliminary findings suggest that in the period from September 2021 on about 2.6% of all food prices have changed on a given day when sales are included and 0.5% when sales are excluded. Across all product groups, the frequency of price changes is considerably higher in the high- inflation period (Sept. 2021 to August 2022) than before. In contrast, the average size of price changes has remained broadly stable over time such that we conclude that currently high inflation is driven by an increase in the frequency of price changes rather than its size.

We analyze to what extent nowcasts of the inflation rate on various COICOP levels can be improved by using web scraped data. For this purpose, we compare nowcasts using webscraped data with nowcasts from various benchmark time series models. A key aspect using webscraped data for nowcasting is product selection and classification.

Using several millions of daily prices collected over the period 2007-2018 in France, we investigate how gasoline retail prices respond to a common shock on marginal cost (i.e. the wholesale gasoline price quoted on the Rotterdam market). We find that the pass-through is complete: a 1% change in Rotterdam price translates to a change in retail price of 0.8%, in line with the share of the wholesale gasoline in total costs. The adjustment is gradual: the full pass through takes about 3 weeks. In a broad class of sticky price models, the ratio of the kurtosis over the frequency of price changes is shown to be a sufficient statistic for the cumulative impulse response of prices (CIRP) to a nominal shock. We provide evidence that the sufficient statistic prediction holds when we look at how gasoline prices respond to a common cost shock. Relating, at the gas station level, the CIRP to moments of the price change distribution, we find that the CIRP correlates with the ratio of kurtosis over frequency, but also with both frequency and kurtosis taken separately. The sign and the magnitude of the correlations are fully in line with theoretical predictions. We also show that other moments do not correlate with CIRP as robustly as the frequency and the kurtosis.

In a broad class of sticky price models the non-neutrality of nominal shocks is encoded by a simple sufficient statistic: the ratio of the kurtosis of the size-distribution of price changes over the frequency of price changes. We test this theoretical prediction using data for a large number of firms representative of the French economy. We use the micro data to measure the cross sectional moments, including kurtosis and frequency, for about 120 PPI industries and 220 CPI categories. We use a Factor Augmented VAR to measure the sectoral responses to a monetary shock, as summarized by the cumulative impulse response of sectoral prices (CIRP), under three alternative identification schemes. The estimated CIRP correlates with the kurtosis and the frequency consistently with the prediction of the theory - i.e. they enter the relationship as a ratio. The analysis also shows that other moments not suggested by the theory, such as the mean, standard deviation and skewness of the size-distribution of price changes, are not correlated with the CIRP. Several robustness checks are explored.

Using micro price data underlying the Harmonized Index of Consumer Prices in France, Germany and Italy, we estimate relative price trends over the product life cycle and show that minimizing price and mark-up distortions in the presence of these trends requires targeting a significantly positive inflation target. Relative price trends shift the optimal inflation target up from a level of zero percent, as suggested by the standard sticky price literature, to a range of 1.1%- 2.1% in France, 1.2%-2.0% in Germany, 0.8%-1.0% in Italy, and 1.1-1.7% in the Euro Area (three country average). Differences across countries emerge due to systematic differences in the strength of relative price trends. Other considerations not taken into account in the present paper may push up the optimal inflation targets further. The welfare costs associated with targeting zero inflation turn out to be substantial and range between 2.1% and 4.5% of consumption in present-value terms.

We utilize a unique micro price data set for Greece that underpins the Greek CPI. It spans almost two decades, during which Greece suffered a large economic shock. We find that during this time there were significant changes in the pricing behaviour of Greek firms. We also find macro-economic developments such as annual inflation and output growth are important factors in determining the frequency and size of price changes. Utilizing the results of the estimations, we set up a small simulation where by the inflation response reacts endogenously via an asymmetric impact on the frequency of price increases and the frequency of price decreases. The results of the simulations capture the Greek inflation developments well. Moreover, they also capture developments in the frequency of price increases and decreases seen in other economies and over different time-periods.

We provide new evidence on the inflationary effects of Quantitative Easing (QE) using Swedish administrative data at the bank, firm, and product level. For identification, we rely on bank-firm lending relationships and the heterogeneous participation rates of banks in the government bond purchase program. Our results show that the bond purchase program led to a significant and persistent increase in producer price inflation. Importantly, we find that the degree of financial frictions considerably influences firms' price response. Low leverage firms do not change their prices by a significant amount, whereas high leverage firms raise their prices significantly. This leverage-dependent price response does not prevail for exogenous changes in the repo rate implying that the transmission mechanism of QE differs from the one of conventional interest rate policy. Our estimates suggest that a bond purchases program of 1.3% of GDP induces a similar-sized price increase than an exogenous fall in the repo rate by 25 basis points.

In a broad class of sticky price models the non-neutrality of nominal shocks is encoded by a simple sufficient statistic: the ratio of the kurtosis of the size-distribution of price changes over the frequency of price changes. We test this theoretical prediction using data for a large number of firms representative of the French economy. We use the micro data to measure the cross sectional moments, including kurtosis and frequency, for about 120 PPI industries and 220 CPI categories. We use a Factor Augmented VAR to measure the sectoral responses to a monetary shock, as summarized by the cumulative impulse response of sectoral prices (CIRP), under three alternative identification schemes. The estimated CIRP correlates with the kurtosis and the frequency consistently with the prediction of the theory - i.e. they enter the relationship as a ratio. The analysis also shows that other moments not suggested by the theory, such as the mean, standard deviation and skewness of the size-distribution of price changes, are not correlated with the CIRP. Several robustness checks are explored.

We use microdata to estimate the strength of price selection – a key metric for the effect of monetary policy on the real economy. We find that price adjustment pressure at the product level does not significantly influence the probability of price adjustment in response to identified monetary and credit shocks, suggesting price selection is absent. This happens even though prices do respond significantly both to aggregate shocks and product-level adjustment pressure directly. Our results are broadly consistent with second-generation state-dependent pricing models and sizable effects of monetary policy on the real economy.

Using CPI micro data for 11 euro area countries covering about 60% of the euro area consumption basket over the period 2010-2019, we document new findings on consumer price rigidity in the euro area: (i) each month on average 12.3% of prices change, which compares with 19.3% in the United States; when we exclude price changes due to sales, however, the proportion of prices adjusted each month is 8.5% in the euro area versus 10% in the United States; (ii) differences in price rigidity are rather limited across euro area countries but much larger across sectors; (iii) the median price increase (resp. decrease) is 9.6% (13%) when including sales and 6.7% (8.7%) when excluding sales; cross-country heterogeneity is more pronounced for the size than for the frequency of price changes; (iv) the distribution of price changes is highly dispersed: 14% of price changes in absolute values are lower than 2% whereas 10% are above 20%; (v) the overall frequency of price changes does not change much with inflation and does not react much to aggregate shocks; (vi) changes in inflation are mostly driven by movements in the overall size; when decomposing the overall size, changes in the share of price increases among all changes matter more than movements in the size of price increases or the size of price decreases. These findings are consistent with the predictions of a menu cost model in a low inflation environment where idiosyncratic shocks are a more relevant driver of price adjustment than aggregate shocks. 

For central banks, nowcasting plays a crucial role in tracking inflation dynamics in real time, particularly in times of high economic uncertainty such as the recent coronavirus (COVID-19) pandemic. In light of the rising market share of e-commerce, a whole universe of digital information is provided by (web-scraped) online prices, which can be tracked at a daily (or even higher) frequency, in contrast to the monthly frequency of the traditional consumer price indices compiled by national statistical institutes. The question is whether this daily information can help in nowcasting the inflation rate. This paper applies correlation analysis and a MIDAS regression approach to assess the forecasting properties of daily online price data for Germany and the euro area.

We look at the additional advantage that mixing macrodata with very granular high-frequency data brings to forecasting inflation at various horizons.

In 2005 the Deutsche Bundesbank presented first results on price flexibility and inflation persistence in Germany based on CPI micro price data. Using a novel database covering more than 80% of the CPI and almost the entire period since the introduction of the euro, we take a new look at price setting in Germany. The depth of the dataset allows us to investigate structural changes in price setting over time, looking at pricing strategies across different types of goods as well as different outlet types, such as supermarkets, discounters and online shops as well as comparing different phases of inflation regimes.

This paper studies price adjustment in a novel monthly dataset of individual product prices of multiproduct firms, merged with firm-level balance sheet and cost data. According to theoretical literature on price setting, the interdependence between the decision on whether or not to change prices (the extensive margin) and the actual amount by which prices change (the intensive margin) contributes to determining the real effects of monetary policy. We estimate the adjustment of prices to shocks to firm-level import costs and energy costs (due to oil supply shocks) along extensive and intensive margins, modelling them jointly to address endogenous selection bias due to state-dependent pricing.

We estimate the response of prices to changes in corporate tax rates (pass-through). For this, we use variations in corporate tax rates across time and location. In Germany, corporate tax rates are set at the municipal level once a year. Each year between 2012 and 2017 about 13% of all municipalities changed their local tax rate. We leverage this fine-grained local variation in tax rates to provide estimates of pass-through that control for macroeconomic conditions. Our study is similar to Baker, Sun and Yanellis (2018), who looked at the variation in tax rates across US states. We find that the pass-through of corporate taxes to prices is about 50%. However, there is significant heterogeneity: in markets with high market concentration, pass-through tends to be 100%, while it is not significantly different from zero in less concentrated markets.

We look at price dispersion across German supermarkets at the individual product level and relate it to goods and local characteristics. We then analyse whether the pass-through of idiosyncratic and common shocks is affected by the uniformity or non-uniformity of prices.

We propose a theoretical model with endogenous quality adjustment which nests the canonical New Keynesian model. In this framework, endogenous quality choice implies a smaller slope than the traditional New Keynesian Phillips curve and amplifies the economy’s response to productivity shocks. In a positive sense, this leads to less reactionary monetary policy where model misspecification matters more than data with poor quality adjustments. In a normative sense, optimal monetary policy is unchanged, as these extensions to the standard New Keynesian model preserve divine coincidence.

This paper models the adjustment of regular and sales prices in retail microdata under the assumption that precise decision-making is costly. We build on two related approaches: “control costs” and “rational inattention”. We adapt and extend the equivalence results of Steiner et al. (2017) to the retail pricing context in order to build a tractable control cost model that approximates the solution of the rational inattention model. We calibrate the approximate rational inattention model to microdata from supermarkets to study retail price dynamics. Using our simulated model to diagnose the roles of stochastic price discrimination versus the inherent discreteness of the rational inattention solution as drivers of retail prices, we find that the former helps explain the “jumpiness” of retail sales, while the latter helps explain the “stickiness” of nominal price points. Our model also suggests that limited memory, in addition to limited information processing capability, may play a role in nominal rigidity.

This paper examines the nature, evolution and sources of price dispersion within European countries. We compare price dispersion across a large set of European countries and explain their differences.

This paper studies the nature, evolution and sources of inflation differences between households in Europe. We compare this inflation heterogeneity with recent findings for the United States (Kaplan and Schulhofer-Wohl, 2017).