The IMF 40 Years Interview Part II: Demand and Supply Across Countries

This post is a continuation of the long version of my interview with Hites Ahir...

 3.      In your research, you have found that demand patterns are consistent across countries, but supply patterns are not. What explains this?

Housing demand is determined by, inter alia, income, prices, demographics, mortgage markets, and of course tastes or preferences.  Let’s focus on incomes today, using Exhibit 1, taken from Malpezzi and Mayo (1987).  These are called Engel curves, after the German economist who began the study of such budget shares in 1857. The horizontal line represents monthly household income in dollars. The vertical line represents the rent-to-income ratio, or budget share, at each income level. Based on analysis of household surveys, we can trace out the average rent-to-income ratio within each city at different levels of household income. Negative sloping lines indicate that rents rise more slowly than incomes (housing expenditure is income inelastic, in economist’s jargon).  A flat line would indicate a constant rent-to-income ratio for rich and poor alike; and a rising line indicates elastic demand. 

Exhibit 1

Demand for housing within a market has been found to be “normal” which means that as incomes rise people consume more housing. No surprise there. But while expenditure on rents rises with income, these expenditures don’t generally rise as fast as incomes rise. Housing is inelastic with respect to income, or in other words a necessity.

Exhibit 1 is based on a classic study Steve Mayo and I carried out years ago in 14 cities from a range of developing countries.  For clarity the dotted lines show only 4 of the 14 city results, although the other 10 were qualitatively similar.  Within markets (cities), housing consumption increases with income, but less slowly than income.  That’s not a surprise. We see this behavior with a number of goods and services. Food is the obvious example – because goods like food and housing are necessities.  

The solid line is different.  The 14 cities all had different average incomes, of course, and so we also plotted the 14 rent-to-income ratios at each city’s average income.  The solid line is the regression line through these 14 points; this shows how the city curves shift up slowly as the entire city’s income level rises.  In other words, over the very long run, as cities develop, elasticities (responsiveness) will tend to be higher than within a single city’s cross section, which is just a snapshot of a particular period.  Housing markets take significant time to adjust, and single cross section snapshots do not reveal truly long run behavior.  

Owner results are qualitatively similar, although the budget shares tend to be a little higher for owners in a given city at a given income level.  

Price elasticities (not shown on the Exhibit) are less well studied, but usually similar in absolute value, i.e. often range from -0.4 to -0.8.  As housing prices rise, the quantity of housing consumed falls; but a bit less rapidly than the prices rise.  Taking these two effects together, if housing’s price elasticity is negative but less than one, as housing prices rise, expenditure (price times quantity) rises, holding income and other things equal.

One immediate consequence of this (and a plethora of other confirmatory research) is that the common practice of adopting a single expenditure ratio as a threshold for so-called housing affordability is very much at odds with actual household behavior. Households around the world treat housing as a necessity; the fraction of income devoted to it systematically falls as incomes rise within a city.

Future demand research will not only upgrade and update these demand estimates, but will expand our focus from developing countries alone to a broader range of emerging markets and developing countries. How these Engel curves behave throughout the entire range of incomes found at different levels of development, and how they change over time, is ripe for further study.

The bottom line about demand is not that it’s identical in all cities.  Other things matter, some readily measured like incomes, prices, and demographics; some are harder to measure including cultural effects on preferences. Nevertheless housing demand appears to be remarkably systematic and predictable, given a few key variables.  Supply is more complicated.

Exhibit 2, from a review I’ve just begun with Lingxiao Li, presents stylized supply curves for housing generated from the elasticities estimated by 6 studies of 15 different countries.  There are many caveats about the strict comparability of these different studies, so think of Exhibit 2 as suggestive, not precise. Housing supply elasticities (responsiveness) appear to be all over the map.  In cities like Bangkok and Atlanta, during the periods studied, small increases in housing prices begin to quickly call forth new supply. That supply can be from new construction; but also from upgrading and modification of the existing stock, through what is often called “filtering.” With these sources of supply working well, supply curves in these markets are flat.

At the other extreme, cities like Seoul and San Francisco have low estimated price elasticities of supply.  In these cities real estate developers and managers of existing real estate assets are constrained in some fashion, and slow to respond to initial price signals. Thus their supply curves are relatively steep.

The estimations behind Exhibit 2 vary in their econometric technique and type and duration of data, so, as already noted, precise numerical comparisons should be taken with grains of salt. But qualitatively, nobody familiar with those cities would be surprised that Atlanta’s real estate supply is much more responsive than San Francisco’s; or that Bangkok’s is more responsive than Seoul’s. Especially in the 1980s when the first studies of these elasticities were completed.

In fact, Bangkok’s real estate supply system was later disrupted by the excesses of the hot money-driven real estate bubble that was the trigger for the 1997 Asian Financial Crisis (see Mera and Renaud). Seoul’s  supply system, while still problematic in some respects, is somewhat more responsive since 1990s era reforms (see Kim and Park). So there’s probably been some convergence in these supply elasticities although we’d certainly benefit from some careful research on this question. A lot of research remains to be done on the supply side, including understanding the time varying nature of changing responsiveness. Mayo and Shepard provide a template for research on this question, and some evidence from Malaysia.

On the other hand many observers argue that the difference between supply conditions in U.S. metropolitan areas has increased over time (Ganong and Shoag).  Certainly the volatility in U.S. housing prices has increased, and also the correlations across markets has also increased, making it increasingly difficult to diversify across regional housing markets (Malpezzi forthcoming).

Malpezzi and Wachter make another salient point about supply. As Samuelson’s famous paper on the Le Chatelier principle reminds us that as we take longer and longer run perspectives, elasticities necessarily increase in absolute value. The central insight is a simple one: the longer the time period under consideration, the more ability we have to adjust either demand or supply or both. As Malpezzi and Wachter showed with simple simulations, and Wheaton demonstrated with a more sophisticated model some years earlier, this implies that markets with an initially more inelastic market much more readily generate large boom and bust cycles. In elastic markets, like Atlanta, and Bangkok, booms are harder (though not impossible) to start. As I have told my students for years, every real estate boom ends with something else that starts with the letter B.

So far we discussed differences in supply elasticities, but not how or why they arise. When we examine these elasticities in terms of deeper determinants, it turns out they are not correlated with income or city size; but neither are they truly random.  Research across countries and across cities demonstrates that much of the variation is explained by supply side constraints, both natural (physical geography) and man-made (land use and development regulations). We have to live with geography, but we can change regulations.  This leads to another central theme of housing research: Regulations have costs and benefits.  Review both carefully.  I see questions below give me a chance to discuss the latter a little further.

References

Chai, Andreas, and Alessio Moneta. "Retrospectives: Engel Curves." Journal of Economic Perspectives 24, no. 1 (2010): 225-240.

Ganong, Peter, and Daniel Shoag. "Why Has Regional Income Convergence in the Us Declined?",  (2013).

Kim, Kyung Hwan, Sock-Yong Phang, and Susan Wachter. "Supply Elasticity of Housing." 66-74: International Encyclopaedia of Housing and Home, 2012.

Kim, Kyung-Hwan, and Miseon Park. "Housing Policies in the Republic of Korea." THE HOUSING CHALLENGE IN EMERGING ASIA,  (2016): 92.

Malpezzi, Stephen. "Economic Analysis of Housing Markets in Developing and Transition Economies." In Handbook of Regional and Urban Economics. Volume 3. Applied Urban Economics, edited by Paul Cheshire and Edwin S. Mills, 1791-1864. New York and Oxford: Elsevier Science North-Holland, 1999.

________. "Residential Real Estate in the U.S. Financial Crisis, the Great Recession, and Their Aftermath." Taiwan Economic Review,  (Forthcoming).

Malpezzi, Stephen, and Stephen K. Mayo. "The Demand for Housing in Developing Countries: Empirical Estimates from Household Data." Economic Development and Cultural Change 35, no. 4 (1987): 687-721.

Malpezzi, Stephen, Stephen K Mayo, and David J) (with Gross. Housing Demand in Developing Countries. Washington, D.C.: World Bank, 1985.

Malpezzi, Stephen, and Susan M. Wachter. "The Role of Speculation in Real Estate Cycles." Journal of Real Estate Literature 13, no. 2 (2005): 143-64.

Mayo, Stephen, and Stephen Sheppard. "Housing Supply and the Effects of Stochastic Development Control." Journal of Housing Economics 10, no. 2 (2001): 109-128.

Mera, Koichi, and Bertrand Renaud. Asia's Financial Crisis and the Role of Real Estate: ME Sharpe Inc, 2000.

Samuelson, Paul A. "The Le Chatelier Principle in Linear Programming." Rand Corporation, memorandum, 1949.

Wheaton, William C. "Real Estate" Cycles": Some Fundamentals." Real Estate Economics 27, no. 2 (1999): 209-211.


Next Post: Progress and Challenges in Developing Country Housing Markets