The objective of the proposed research is to facilitate the evolution of macroeconomics from reliance on the National Income and Product Accounts (NIPA) toward an internationally integrated system of City Income and Product Accounts (CIPA). Much is gained by this transition in terms of the scope of economic research and policy analysis while little is lost. That little is lost is evident in the fact that world urbanization increased from 13% in 1900 to 49% in 2005 and is predicted to rise to 60% by 2030 (2006 UN World Urbanization Prospects report). What is gained is substantial based on preliminary results emerging from a number of sub-fields of economics. International economists using city-level price surveys, including work by the co-PIs and collaborators, find that many facts arising from studies of national price indices fail to emerge when city level data is employed. Price adjustment is much faster than what the national data show. Goods that are less traded have larger and more persistent deviations than goods that are more traded. Market segmentation arising from national borders is less severe when absolute LOP deviations are the metric than when time series volatility of aggregated CPI indices are the metric. Three core themes surface in the context of these facts. First, the service sector looms large, both in the distribution of traded goods to final consumers in retail markets and in local services production (education, medicine, emergency services, utilities and transportation infrastructure). Second, markups of price over marginal cost play a central role in accounting for geographic price dispersion. Third, two thirds of international trade has a single multinational firm at one end of the transaction. While this suggests levels of industry concentration conducive to geographic price discrimination, there is still much to learn about the relationship between the geography of markups and the geographic concentration of product-level production. Intellectual Merit The intellectual merit of the research is to provide data infrastructure to support Dynamic Stochastic General Equilibrium (DSGE) modeling at the microeconomic level. This allows integrated study of heterogenous economic responses to macroeconomic shocks as well as the spatial scale of microeconomic spillovers. LOP is a compelling reason to study goods markets and cities in a general equilibrium framework. The challenge is to a build model that incorporates LOP deviations in a realistic fashion. Some markets are geographically segmented (medical services, public education), others are globally integrated (agricultural, manufactured goods). The equilibrium of a city depends on the balance of segmentation across the goods, labor and capital markets. Broader Impact The broader impact is a novel way of thinking about spatial economic interaction. Shocks and policies need not be defined on the basis of arbitrary political divisions. Cities are microcosms of the world economy that offer insights into the social sciences. Since services are disproportionately publicly provided, cities provide laboratories to study political economy, including issues that span national boundaries. Linking the data and models developed here with Google Earth, students and teachers will be able to explore economic, geographic, cultural and political dimensions of cities in a more integrated and rigorous manner.
The research program seeks to understand the sources of price differences across cities of the world. This involves three complementary research activities. The first activity is the creation of data archives for use by the public and the research community containing absolute prices of individual goods and services at the city level. The second activity involves summarizing the prices across citiesand over time using formal statistical analysis. The third activity involves building economic models to deepen our understanding of the sources and consequences of international price differences. Figure 1 emphasizes differences across goods, the upper panels show the relative prices of apples and haircuts across U.S. and Canadian cities from 1990 to 2005. Each line is the log relative price across a city pair, thus a value of 0.50 is approximately a 50% price difference. If the LOP held for all time periods and city pairs, all the lines would fall on the horizontal 0 line. In fact, what we see is that deviations from the LOP are typical of both apples and haircuts (chosen as examples of traded and non-traded services, respectively). The data panels used in our research typically involve hundreds of such goods and services. The total variance of LOP is almost the same for these two items, which is surprising as haircut markets are obviously more geographically segmented. It turns out that this segmentation is evident in the middle panel which contains the long-run mean price differences. It appears, then, that shocks to the apple market give rise to time series fluctuations despite world-wide integration (as seen in the last panel which are the deviations of LOP from the long-run means). This analysis points to the need for models to capture both the sources of long-run price differences and time series fluctuations around those differences. Figure 2 focuses on location differences. It contains the relative prices of a randomly selected sample of goods for: Detroit, Pittsburgh and Bangkok. The left panel presents relative prices of goods across the two U.S. cities, the right panel is the price in Detroit relative to Bangkok. Thailand experienced a currency crisis and massive depreciation of its nominal currency late in 1997. We see this had the expected effect of lowering prices in Detroit relative to Bangkok (the lines fall in level post 1998). The figure also reveals that much of the variation is specific to the good, with the exception of the dramatic currency crisis event itself (the sharp downward spike). The lower panel shows the real devaluation is part of a longer run trend in which prices are rising in Bangkok toward levels in Detroit, consistent with a process of ongoing development. This figure shows that while the price level may be affected in an obvious fashion by large nominal exchange rate changes, there are many idiosyncratic price trends within the distribution that may be caused by changes in productivity or other microeconomic supply and demand shifts. Such shifts alter international relative prices and often it takes a non-trivial amount of time for world markets to reallocated goods and inputs in response to these changes. The data in these figures are from the Economist Intelligence Unit and were chosen because they demonstrate key facets of cross-sectional and time series variation of international relative prices that we see in our public archives. The public archives available at the Centers for International Price Research homepage: www.vanderbilt.edu/econ/cipr/# include: Commodity prices (about 549 different commodities) for Boston, Charleston, Cinncinati, New York, Boston, New Orleans, and Philadelphia, monthly from 1700 to 1861. Retail prices of hundreds of goods and services for 13 (mostly capital) cities in Europe, 1975, 1980, 1985, 1990. Retail prices of almost 700 goods and services across 71 Japanese cities, monthly 2000-2006. Retail prices of 223 goods and services in 12 Ecuadorian cities, monthly from 1997 to 2003. Model mechanisms we have emphasized to account for the long-run and time series deviation are the following: Trade costs and distribution margins (retail costs and markups by retails) seem to contribute about equally to long-run price deviations at the level of individual goods and services. Price deviations in goods with more local cost content (i.e., retail goods with higher distribution costs) tend to be determined by local conditions in each market whereas prices of traded goods tend to deviate by a traditional shipping cost. When micro-data are aggregated, distribution costs and markups tend to dominate as sources of international price dispersion whereas traditional trade costs tend to average out across goods. Time series variation in LOP seem to be dominated by shocks that are idiosyncratic to the good or sector, such as changes in productivity in contrast to price level deviations when seem to be more dependent on nominal exchange rate movements and the overall state of the macroeconomy.
