Ebook Menu Costs, Multi-Product Firms, and Aggregate Fluctuations
It has been well documented, using both survey evidence, but also direct observation, that individual goods prices are sticky. The latest major piece of evidence supporting the notion that prices adjust sluggishly is a study by Bils and Klenow (2004) who find, based on a dataset of prices collected by the BLS, that half of the consumer goods prices in the US economy adjust less frequently than every 4.3 months. Whether these firm-level rigidities have important macroeconomic implications is however still an open question.
Most recent work analyzing the consequences of nominal rigidities assumes that firms employ ad-hoc policy rules and does not explicitly model the source of price stickiness. These, time dependent models postulate that the timing of price changes is exogenous and unresponsive to the state of the world. Information-gathering costs or institutional restrictions are presumed to give rise to this behavior, but these frictions are, with a few exceptions, rarely modeled. The fact that these models lack micro-foundations makes them inappropriate for the study of many interesting policy questions, but also reduces the number of dimensions along which the theory can be tested.
Recently the profession has witnessed a growing interest in an alternative class of models in which agents solve fully-specified problems and nominal rigidities arise endogenously, due to fixed physical (menu) costs of changing prices. These, state-dependent or (S, s) pricing models can be traced back to the work of Barro (1972) and Sheshinski and Weiss (1977, 1983), but only recent advances in numerical solution techniques have enabled researchers to study dynamic, general equilibrium versions of these economies. Their aggregate implications are, nevertheless, not well understood. In particular, the ability of firm-level nominal rigidities to generate business cycle fluctuations from nominal shocks crucially depends in these models on the distributional assumptions made to aggregate the economy.
The predictions of menu-cost models range from stark neutrality to cases in which the economy is virtually indistinguishable from time-dependent setups. Golosov and Lucas (2004) study the properties of a model with firm-level disturbances capable to match the fact that the magnitude of price changes is large in the US economy: 10% on average, much larger than what can be explained by aggregate shocks alone. They find that the model produces very little output volatility from monetary shocks, a result similar in spirit to that of Caplin and Spulber (1987). Klenow and Kryvtsov (2004) reach an opposite conclusion. They document that there is little evidence of across-firm synchronization in the US price data, contrary to what standard menu cost models predict. They find that a model with time-varying costs of price adjustment that can replicate this feature of the data behaves identically to a time-dependent sticky price model and produces large output variability from monetary shocks.
This paper revisits the question of whether menu costs of price adjustment can, in fact, generate a monetary transmission mechanism. I argue that a truly micro-founded model must be rendered consistent with the price adjustment practices observed at the firm level before one can proceed to study its aggregate predictions. I start by documenting several salient micro-economic features that characterize firm pricing behavior. To this end, I employ a large set of scanner price data collected in a number of grocery stores over a twelve-year period. In addition to the large frequency and magnitude of price changes, documented by Klenow and Kryvtsov (2004), I document two additional features of the data. First, a large number of non-zero price changes are small in absolute value. Second, the distribution of price changes, conditional on adjustment, exhibits excess kurtosis.
Download
PDF Ebook Menu Costs, Multi-Product Firms, and Aggregate Fluctuations
Posted in :
