قیمت دارایی واقعی در پکن, 1840-1644

This paper provides the first estimates of housing price movements for Beijing in late pre-modern China. We hand-collect from archival sources transaction prices and other house attribute information from the 498 surviving house sale contracts for Beijing during the first two centuries of the Qing Dynasty (1644–1840), a long period without major wars, political turmoil, or significant institutional change in the Chinese capital. We use hedonic methods to construct a real estate price index for Beijing for the period. The regression analysis explains a major proportion of the variance of housing prices. We find that house prices grew steadily for the first half-century of the Qing Dynasty and declined afterwards in both nominal and real terms through the late eighteenth century. Nominal prices grew starting in the late eighteenth century and declined from the early nineteenth century through 1840. But these price changes occurred with contemporaneous price changes in basic measures of the cost of living: there was little change in real terms to the end of our period

In this paper, we assemble a new and unique dataset for house prices in Beijing between 1644 (the start of the Qing Dynasty) and 1840, a period without war, major political turmoil, or significant institutional change in the Chinese capital.3 We use the dataset to document long-run house price trends in Beijing. We combine this long-run house price index with historical information on population, economic conditions, government policies, and natural, political and economic shocks, and discuss what the driving forces behind the long-run historical price changes might be. We combine the historical house price series for Beijing with a series for consumer goods prices and wages to estimate the change in real house prices. More speculatively, we relate these to recent and present price levels for housing in Beijing. The contracts underlying our dataset have not been previously used by economists. In fact, scholarly research on the historical housing market in China has been quite limited.4 Lack of data hampers solid inquiry into this unknown area of research. Our work on historical housing prices in Beijing is the first quantitative study of China's historical housing market. This paper makes several contributions. First, the dataset we have created covers a long span of the Qing Dynasty period, from its rise through its heyday and into its decline. The data are uniquely valuable: the house transaction contracts from which our dataset is drawn are the only reliable primary source for information on the development of the housing market in Qing Dynasty China.5 Besides price, we also collect characteristics for properties that are the subject of transaction, and we identify the location of these properties. We are then able to link the locations to centers of political and commercial activity in historical Beijing. Second, by estimating a hedonic price model, we construct a time series of housing prices for this period of nearly 200 years, as well as other economic series. Moreover, we construct other economic series, including measures of the price of rice and of consumer prices more broadly over the period, which we use to deflate the real estate price index to create a series in real rather than merely nominal terms. The hedonic model provides insights on the value to be attributed to particular characteristics of housing in this historical period, including the important factor of location. The housing price index provides insights on the time path of housing prices over time. The latter can be compared to other historical trends, for example those of prices more generally. Third, in recent years, much work has been published that aims to compare real income and standard of living in China with those in other major economies in a longer historical horizon. Concerted efforts in collecting more historical price and wage data from Chinese historical archives have produced academic achievements.6 Such studies provide valuable insights for us to understand the long-run causes, processes and consequences of the divergence between Europe and Asia. Housing-related expenses are an important item of household consumption, and changes in house prices and rental costs have large impacts on people's real income and standard of living. Due to data limitations, however, there is no study so far that incorporates housing expenses into the consumption basket. Our estimation of the long-run house price index and house values provides a benchmark that scholars can use to estimate expenditures on housing for Beijing residents in the Qing Dynasty, thus allowing more reliable estimates of costs of living and standard of living in that period in China. Many scholars have, of course, studied housing prices, indeed for many places, using hedonic methods. Such studies covering short intervals of time are common.7 Long-period studies, in contrast, are very few in number.8 One which comes close is the study of Case and Shiller (1987). The Case/Shiller index, in widespread use in the U.S., is based on repeat sales of the same property and starts in the late 1980s.9Shiller (2005) combines these data with other long term data series, including construction costs, to create series beginning in 1890. Linking these indexes, Shiller finds that until the recent run-up of prices in the U.S. starting in the mid-1990s, housing prices adjusted for inflation were basically constant. The only other study that we are aware of that covers a long historical period, and the only study that covers a period comparable to ours, is Eichholtz (1997). That paper introduces a biennial historic index of real estate values for the period 1628 through 1973 for Amsterdam based on the transactions of the buildings on the Herengracht, one of the main canals in the city and finds, similarly to Case, that housing prices essentially do not change much when viewed over extended periods. Our resources differ from those of Eichholtz in two respects. First, our index is temporally much coarser due to the relative paucity of reported transactions which we have access to in this period. Second, we are not able to track repeated sales. Instead, we construct a hedonic index with the usual dummying strategy for time.10 Controlling for a variety of property characteristics including location, we construct a price index from the coefficients on the time variable, as is standard in the hedonic price index literature. Measuring prices for same-house sales over time almost perfectly controls for house characteristics. In the absence of such rich information, the standard method for constructing house price indices involves including house characteristics directly in regressions and estimating price indexes over time from the coefficients of dummy variables for time. This method has been shown to produce robust price indices. But we believe that estimation of such an equation for a 200 year period has not been attempted before now. The equations perform well, in part due to the availability over time of data on a number of attributes. The estimates throw light on the importance of these property characteristics, including size of property, its location, the availability of onsite water supply, what building materials are used, and whether or not there had been formal legal registration of the property sale. The paper is organized as follows. Section 2 introduces the historical background of Beijing's house market during our sample period. Section 3 explains in detail how we extracted variables and processed the data from original contracts. Section 4 explains how, using hedonic methods, we construct a house price index for Beijing from these data. Section 5 shows several robustness checks we make for our hedonic regressions. Section 6 compares our house price index with price indices for other consumption goods and wages. Section 7 offers a brief conclusion.

With 498 surviving real estate transactions contracts, we use hedonic regression methods to estimate a long-run house price index for Beijing between 1645 and 1840. We are able to explain a large amount of the variation in housing prices by the characteristics of the properties, including the number of rooms and location. We also create rice- and more broadly based consumer price indices. We compare the house price index results with these series and estimate real house price series. We find that real house prices increased in the second half of the seventeenth century, declined in the first half of the eighteenth century, and then remained relatively constant in real terms. We are in this study limited in our data over this approximately 200 year period and thus we estimate values at intervals of 25 years. Nonetheless the observations we do have for the second half of the seventeenth century indicate that price increases are consistent with the rapidly growing economy and urbanization that historians have attributed to the period. The decline of house prices that we observe in the first half of the eighteenth century is consistent with a government policy change that allowed the construction of new homes and deregulated housing markets. We briefly compare house prices in the Qing Dynasty with prices in 2002 and in 2012. It is well known that house prices increased dramatically between 2002 and 2011. Even in 2002, prior to the run-up, we see that prices appear to have increased between 1840 and 2002. While this increase raises questions about more recent price trends, our results for the period 1644 to 1840 are consistent with the findings in Shiller (2005) and Eichholtz (1997) which show no significant price increase in real terms, for 150 years in Shiller and for nearly 350 years in Eichholtz.