VoxEU Column Monetary Policy

House and goods price inflation: Evidence from bar code data

Rising prices have long been a concern of monetary policymakers due to wealth effects on spending. This column presents evidence that local demand effects from  house price increases result in significant local price inflation. Households living in locations with rapidly increasing real estate prices will also face rapidly increasing costs of goods purchased in local stores.

What are the macroeconomic effects of changing house prices?  The Great Recession has led to a flurry of interest in this question.  A number of recent papers have argued that house price-induced changes in wealth lead to large changes in household spending and were a significant force in the recession (Mian et al. 2013, Case et al. 2013, Mian and Sufi 2014).

In recent work, we show that in addition to these spending effects, house price changes also lead to significant changes in retail price inflation which have important implications for monetary policy (Stroebel and Vavra 2014).  Using barcode level price data from 2,400 US zip codes on a variety of products sold in grocery and drug stores, we estimate that when local house prices double, product prices will increase by 15 to 20 percent. That is, households living in locations with rapidly increasing real estate prices will also face rapidly increasing costs of goods purchased in local stores.

The left panel of Figure 1 shows that retail-price inflation was highest in locations with high housing price growth during the housing boom.  Conversely, the right panel of Figure 1 shows that the regions with the biggest declines in house prices during the housing bust experienced the lowest inflation over this time period.

Figure 1. Retail-price level (MSA) time series

Note: Figure shows the average retail price level over time for MSAs in the top quintile (solid black line) and bottom quintile (dashed blue line) of house price appreciation for the period 2001-2006 (Panel A), and the period 2007-2011 (Panel B).

Why do increases in house prices lead to increases in retail prices?  By definition, an item’s price is always equal to its cost plus some markup.  This implies that changes in retail prices always reflect either a change in costs or a change in markups.  While either channel would be interesting, we argue that our empirical relationship is primarily driven by markups.  To show this, we first control for various observable costs such as local wages, commercial rents, and wholesale costs to show that they cannot explain why retail prices rise with house prices.  If costs do not explain our result, then it must be explained by variation in markups.  But why should retailers increase markups when house prices rise?  We provide evidence that when house prices rise, this increases homeowners’ wealth.  As homeowners feel wealthier, they pay less attention to the particular prices they pay for retail goods.  Retailers then respond by increasing their markups.

Owning versus renting

We provide two pieces of evidence that this housing wealth effect on markups drives our results.

First, we show that there is a strong interaction between house price growth and retail price growth.  Zip codes with high concentrations of homeownership exhibit a much stronger relationship between retail prices and house prices.

Figure 2. Retail-price level (Zip code) time series

Note: Figure shows the average retail price level over time for zip codes in the top quartile (solid black line) and bottom quartile (dashed blue line) of house price appreciation in the US between 2001 and 2011. Panel A shows results of zip codes in the bottom quartile of homeownership rate, Panel B shows results of zip codes in the top quartile of homeownership rate.

In fact, in zip codes which consist primarily of renters, house-price increases do not lead to any increase in retail prices.  While house price increases should lead to the same increase in costs regardless of local homeownership rates, the effects on local wealth will depend greatly on homeownership.  If there are many homeowners in a zip code then house price increases will lead to large increases in wealth and thus large increases in markups.  In contrast, house price increases in zip codes where most households rent do not yield similar increases in local wealth, and that is why we see no increase in markups or retail prices.

Second, we provide direct evidence that homeowners’ retail price sensitivity changes when house prices increase.  Using household shopping data collected by AC Nielsen, we show that when house prices rise, homeowners spend more, but they purchase a smaller share of generic goods, use coupons less intensively, and purchase fewer goods on sale.  In contrast, we see no such effects for renters.

Causal inference

We document a robust positive relationship between house prices and retail prices and argue that this occurs because house price increases cause wealth to rise, which leads firms to raise markups.  It is important to note that correlation does not imply causation.  It is possible that the correlation between retail prices and house prices we find instead reflects some third factor that moves both variables, or reflects reverse causation with increases in retail prices driving increases in house prices.

We provide direct evidence for our causal interpretation in several ways. 

  • First, the interaction we find with local homeownership rates is exactly what would be predicted through a wealth effect channel, while it is hard to explain why some third factor such as local supply shocks would interact with the level of homeownership. 
  • In addition to this evidence, we also use an instrumental variables regression to isolate ‘exogenous’ house price movements that are unrelated to other variables in the economy. 

The appeal of this second technique is that these regression results can be interpreted as causal rather than only as correlations.  In order to try to identify exogenous house price movements, we follow Mian and Sufi and use the local elasticity of housing supply to predict exogenous house price movements.  This supply elasticity measures the presence of mountains, oceans or restrictive zoning that makes it difficult to build housing in certain locations.  In response to a national increase in housing demand, house prices will increase more in locations where it is more difficult to build, inducing exogenous variation in house prices unrelated to other regional characteristics.  This instrumental variable strategy produces even stronger results than using simple regressions and provides strong evidence that house price increases cause retail prices to increase rather than vice versa.

Policy implications

The fact that markups rise with household wealth and demand has important implications for monetary policy.  In standard New Keynesian models, the real effects of monetary policy depend crucially on how inflation responds to changes in demand.  When the Federal Reserve cuts interest rates to stimulate the economy, it increases the amount of money in circulation and increases nominal demand.  By definition, nominal demand is the product of the price level P and real output Y.  Thus, the less inflation responds to monetary stimulus, the greater will be its real effect.  In standard New Keynesian models, increases in nominal demand lead firms’ costs to rise as wages increase and firms reach capacity constraints.  Given the previous identity that price equals cost plus markup, this increase in costs eventually leads prices to rise, and the speed of that price response determines the real effects of monetary stimulus.

In this standard New Keynesian framework, the effect of monetary stimulus on prices works entirely through this cost channel – these models assume that increasing demand increases costs, but that it has no effect on firms’ desired markups. Our evidence suggests this transmission mechanism is incomplete.  When the Fed stimulates the economy it will increase nominal costs, but it will also increase firms’ desired markups.  This increase in markups will amplify upward price pressure and mute the real effects of monetary policy.  In this sense, cyclical shopping behavior and its interaction with firms’ optimal markups creates extra inflation headwind that works against stimulative monetary policy.  Policymakers need to take this into account when trying to forecast the effects of interest rate cuts on real economic activity.

References

Case, K E, J M Quigley, and R J Shiller (2013),  “Wealth Effects Revisted: 1975-2012”, NBER Working Paper No. 18667.

Mian, A, K Rao, and A Sufi (2013), “Housing Balance Sheets, Consumption, and the Economic Slump”,  Quarterly Journal of Economics.

Mian, A and A Sufi (2014), “House Price Gains and US Household Spending from 2002 to 2006”,  Fama-Miller Working Paper.

Stroebel, J and J Vavra (2014), “House Prices, Local Demand and Retail Prices”,  NBER Working Paper No. 20710.

2,625 Reads