Will China Crash in 2015?
by Mason Gaffney
As published in the American Journal for Economics and Sociology

Abstract. Loosely derived from Henry Georgeís theory that land speculation creates boom-bust cycles, a real-assets model of economic crises is developed. In this model, land prices play a central role, and three hypothesized mechanisms are proposed by which swings of land prices affect the entire economy: construction on marginal sites, partial displacement of circulating capital by fixed capital investment, and, finally, the over-leveraging of bank assets. The crisis of 2008 is analyzed in these terms along with other examples of sudden economic contractions in U.S. history, recent European experience, and global examples over the past 20 years. Current conditions in China in 2014 are examined and shown to indicate a likely recession in that country in 2015 because their banks are over-leveraged with large-scale, under-performing real estate loans. Finally, alternative methods of preventing similar crises in the future are explored.

The concern about economic instability is understandably limited in the minds of most people to their own country. Thus, the economic and financial contraction in the United States that began in 2008 has been the object of most interest to American economists. When they make comparisons, the experience of the U.S. in the 1930s tends to be the primary event that is viewed as parallel to the current crisis.

A general theory of the business cycle or economic instability should not be derived from or tested solely by two events. If a theory is valid, it should apply to numerous cases in the past, and ideally, it would serve as the basis for predicting future cycles of boom and bust, growth and decline. We cannot, like scientists in a laboratory, create the conditions we wish to study, but economists can attempt to be more scientific than we have been until now by examining as many cases as possible from many countries and time periods.

This article will begin with a conceptual model that first offers an explanation of numerous panics, crashes, or crises in the history of the United States. It will then discuss its relevance to recent expansion and contractions in several European economies. Finally, it will analyze the prospective problems facing the Chinese economy in the fall of 2014, with the hope of demonstrating the value of this model for prediction, not merely description. By applying the model to a number of different circumstances, I hope to show that it is a universal model, capable of explaining cycles of economic expansion and contraction throughout the world.

The theory presented here derives from a simple model developed by Henry George ([1879] 1979: BK V, Ch. 1). George blamed the periodic paroxysms in modern economies on the effects of land speculation. In his view, if enough people held land off the market in the hope of a price increase, the resulting artificial scarcity of land would raise rents and land prices, driving down wages and returns to productive investment. Eventually this process would reach a limit, land prices would fall, workers would be laid off, and factories would close. We should give George credit for seeing the general outlines of this process, but we must elaborate on his ideas to incorporate elements other than land prices. I have elsewhere presented a thorough explanation of the model developed below (Gaffney 2009), so I shall merely summarize it here.

To make sense of what is happening to China in 2015 and the United States in 2008 and 1927, we must shift our attention away from the details of each crisis and attempt to detect, amid the noise and varying particulars, a general pattern of economic crisis. I will develop here a ìreal assetsî model or theory that will be useful in understanding that pattern. The hypothesis here has four elements:

1) a rise and fall of land prices, resulting mostly from autonomous real economic changes. These are less visible and less measurable than purely monetary and fiscal changes, which may reflect and even reinforce the real changes but not initiate them,
2) investment in projects at the margins, both in terms of geographic location and value,
3) concomitant changes in the structure of capital investment to favor structures with long payout periods, and
4) an increase in bank leverage ratios as a result of lower capital turnover, leaving many banks technically in default by the time the land price bubble bursts.

Land Price Changes
The present hypothesis begins a posteriori from observing land prices increasing over a period of five to eight years after a trough. This contrasts to common scenarios that cast the banking system as the autonomous factor initiating economic crises. Once price increments begin to seem the normal, irrational expectation, staid banks and other financial institutions turn to lending on land collateral, and expand their balance sheets to accommodate the increased investment in real estate. But banks are responding to an external stimulus (an apparent improvement in the value of real estate) rather than creating the conditions for a boom on their own. It is only in the late stages of the land price boom that banks, when they are lending money on increasingly marginal sites, must develop creative accounting methods to circumvent the financial regulations that were put in place during a previous period of contraction to prevent reckless lending.

The first sign of a new cycle of boom and bust is a self-generated rise in land prices, which shows up in the form of higher priced houses. When we speak of an increase in housing prices, what we really mean is the change in the price of urban land on which the houses are built. Since the actual housing stock depreciates over time, it makes no sense to conflate the two. If the price of housing rises 8 percent, the price of land must have increased by, say, 15 or even 20 percent to account for the stability or decline of the portion of value invested in physical capital.

A land price occurs naturally as a result of increased economic productivity or, often enough, a period of peace dividends following a major war. For the first few years after a crash, the land market remains unnoticeable. Buyers are wary of investing in real estate immediately, and banks are even warier of lending for that purpose. In addition, banks are unable to lend much because they have to retrench in order to lower their leverage ratios. Nevertheless, four or five years after a crash, land prices will start creeping back upward as a result of general economic growth. Initially, that will cause an increase in the price of existing houses. (Actually, the price change will represent an increase in the value of sites, not the buildings, which are depreciating.) Increases in real estate prices will signal to home builders and commercial real estate developers that the time has come to build new homes and offices. At that point, perhaps 10 to 12 years after the last crash, a new speculative boom begins. At this point, the rise in land prices accelerates because a rise stimulates more investment, which stimulates a faster price increase, and so on. This occurs first in residential markets, followed closely by commercial real estate. Land price appreciation becomes noticeably higher than alternative investments, and a frenzy takes hold in which more and more people imagine they can make money without effort by investing in real estate. Many of those caught up in a frenzy fancy they are the rational ones, as revealed by the phrases they publish, like Fisher’s “permanently high plateau of prices”, and Sargent’s “rational expectations”, and Will Rogers’ “Buy land, they ain’t makin any more of it!”

The rise in land prices must eventually reach a limit. The ratio of loan repayments to cash flow eventually becomes high enough, even to the point of exceeding unity, to reverse the speculative frenzy and cause a sell-off. There is a plateau at the peak, as owners who bought for short-term gain are now reluctant to take a loss, so as the market softens, bid prices fall faster than asking prices, and sales slow down for a few months.

According to S&P/Case-Shiller Home Price Index (2014), the index of home prices hovered at 206 from June to August 2006 (January 2000 = 100), before beginning a decline to 139 in April 2009. (The price of land is more volatile than home prices, since the relative price stability of the housing stock moderates the combine price of land and housing. Once prices start to fall, some speculators begin selling quickly, accepting declining bids in order to avoid holding property in a falling market. Others, with greater holdout power and more sanguine expectations, settle in for the long wait hoping prices will recover. The price of land of those few parcels that do sell then starts to fall, faster than it rose. Land prices may crash in two or more waves, the housing market leading the commercial market by about two years, as happened 1927-29 and 2006-08.

If there were a national land price index, it would be possible to demonstrate this pattern. In the absence of such an index, we can use data on construction from the United States Census Bureau (2012; 2014a). The value of residential construction rose 95 percent from 2000 (annual average) to March 2006, then declined by 66 percent by February 2011. A similar phenomenon occurred in nonresidential construction, which includes not only commercial, industrial, and service sector construction, but also infrastructure such as roads and sewer lines, harbors and airports. From June 2002 to November 2008, it rose 232 percent, then fell 29 percent by May 2011. There was a lag of 32 months between the peak of housing construction (March 2006) and the peak of non-residential construction (November 2008).

National data on government gross investment in fixed assets (around 20% of private investment) shows much less volatility than private investment, which is surprising, given the importance of local infrastructure investment in fueling a boom and the immediate depreciation of abandoned public infrastructure after a crash. However, the absence of volatility in aggregate data may simply reveal the relatively small number of cities affected by ìirrational exuberanceî that can create a national economic crisis. Thus, national data may obscure what is going on in key local economies. There also temporal idiosyncracies, varying with the ideologies of politicians. Thus the canal boom of 1820-40 was a nationwide mania in spite of President Jacksonís refusal to use Federal funds directly. This is an area for future research.

Note that in the case of housing, the peak of construction occurred three to five months before the peak of the S&P house price index. This would suggest that housing starts might represent one of the best leading indicators we have of an impending contraction in the economy.

The same pattern of successive waves of housing and nonresidential construction cycles can be found in the 1920s, indicating that this model is useful in explaining the Great Depression as well as the most recent crisis. The value of housing construction rose 167 percent from 1921 to 1926, then declined 93 percent by 1933 (United States Census Bureau 1975a: Column N32). Nonresidential construction rose 93 percent from 1921 to $2.7 billion in 1929, then fell 85 percent by 1933 (United States Census Bureau 1975a: Column N36). The peaks of residential and nonresidential construction were thus around three years apart.

Of course, a rise and decline in housing construction is not a perfect proxy for changes in land prices, but it is a good approximation. A large increase in housing construction occurs during the boom phase of the land price cycle because (a) building and selling houses is one way to generate immediate revenue from rising land prices and (b) rising land prices generally encourage investment in capital with low turnover (as will be explained below). Since there are no land price indices in the United States and most other countries, construction data must serve as a proxy in following the economic events that lead up to a financial crisis.

Download the PDF for the full paper