Housing starts are an important element of residential investment, an important component of total economic output. Of course, the supply side of the market responds to demand, so the extent to which the government-sponsored enterprises (GSEs) Fannie Mae and Freddie Mac affect housing should be equally apparent on the supply side, as measured by housing starts.
If housing starts are not responsive to price changes, then other factors may have contributed to regional differences. Indeed, many scholars assert that regulations played an important role in limiting the amount of the land available for residential housing.
For example, Harvard University's Edward Glaeser argues that restricting housing supply through regulations leads to greater volatility in housing prices. He compared the Boston region with the Atlanta region, which issues about seven times as many permits as the Boston region, and found that housing prices in Boston were much higher and more volatile than housing prices in Atlanta. In addition, Atlanta did not share in the Boston Boom of 1980-1988 when prices doubled, and did not suffer from Boston's housing bust of 1988-1994 when prices lost 50%.
Both demand and supply factors may contribute to the story. Indeed, with supply constraints and regulations in place, bubbles can form with a small increase in demand. With an inelastic housing supply (vertical supply curve), a small increase in demand can lead to a huge increase in housing prices. In other words, land-use regulations and geographic land constraints can have important effects on housing booms and busts, but this is not the whole story. As seen in the more detailed econometric results, many other economic fundamentals affect housing starts.
Economists have not reached a consensus on the specification of housing supply models. In earlier studies, housing starts were a function of the level of price and construction costs, but more recent studies link housing starts to changes in housing prices and costs.
On the demand side, the Federal Reserve's policy of maintaining extraordinarily low short-term interest rates, which contributed to lower mortgage interest rates in 2003 and thereafter, intensified the housing price boom, which occurred because of easy access to mortgage credit, among other factors. While the results reported in this paper show that high interest rates may not affect housing starts, it appears that low interest rates, in part, steered individuals to purchase new houses.
Housing starts jumped up by the end of 2003 and remained high until the bubble burst in early 2006. The rise in demand for real estate accelerated housing price inflation, which reached 10% in the fourth quarter of 2004 and remained over 10% for the two consecutive years. With the rise in housing prices, delinquency and foreclosure rates started to fall.
However, as short-term interest rates returned to their initial levels, demand fell sharply and housing prices dropped below their equilibrium levels, bringing down home equities to levels lower than their mortgages. Underpriced home values, along with hikes in delinquency and foreclosure rates, led to the bubble bursting and the meltdown in the housing market in 2007. In fact, both the demand and supply sides of the housing market contributed to the meltdown, which put millions of borrowers upside down with negative equities.
The housing start variable is the main component that determines the net increment to the stock of housing supply. This component, which is withdrawn from the stock of the supply, is called wastage, which refers to ruins, damages or other outflows from the stock. Thus, housing starts, which serve as a proxy for completed new housing units coming onto the market over a given period, can be defined as the change in the housing stock minus the outflow from the stock or wastage.
Housing starts today reflect differences between the future expected stock supply of housing relative to the expected future stock demand for housing. Thus, in an equilibrium condition, housing stock equals expected stock demand.
The mortgage interest rate and housing prices are the most important channels that affect housing starts, particularly during the booms. However, other economic fundamentals also affect home buyers' decisions and play an important role in shaping housing starts.
Theoretically, the quantity of housing starts is positively correlated with mortgage availability, savings rate, real gross domestic product (GDP), price-to-rent ratio, household assets and median home prices, and negatively correlated with mortgage interest rate, Consumer Price Index inflation, household debt and vacancy rate.
Many economists argue that the presence of Fannie Mae and Freddie Mac in the secondary mortgage market affects the primary mortgage market rate. Several econometric studies have found that the interest rates without Fannie and Freddie would be 20 to 35 basis points (bps) higher, but the estimates vary substantially depending on the empirical specification, sample and time period.
Based on these and many other empirical studies, we assume that without Fannie and Freddie, the housing market for 1980 to 2010 would have seen mortgage interest rates average 25 bps higher than what actually prevailed. In fact, this would mean a one-time jump in the mortgage interest rate, which affects the intercept of the regression model, not its slope. Therefore, it is a transitional, short-term effect.
The average elasticity of housing starts to mortgage interest rate at the national level is 0.20. To illustrate what a 25 basis point increase in the mortgage interest rate can do to the housing starts, we can use the elasticity of housing starts to mortgage interest rate, 0.20. Multiplying this elasticity by a 25-basis-point increase in the mortgage interest rate yields a 0.05% reduction in the housing starts. In other words, when compared with changes in other fundamentals, such as real GDP or CPI, a small upward shift in the conventional mortgage interest rate has a minor, negative, one-time effect on housing starts.
As a matter of fact, any small change in real GDP or CPI would affect housing starts much more than any change in mortgage interest rates. For instance, a 2% increase in CPI reduces the housing starts by 8% to 10%, which is substantially larger than the effects of the changes in mortgage interest rates that would result from shutting down Fannie and Freddie.
One of the reasons that some studies have found higher elasticity of housing starts to mortgage interest rate is that they have controlled for only a few – or zero – economic fundamentals in their studies. The results of this paper indicate that the economic fundamentals – such as real GDP, CPI, household assets, household debts, family income and price-to-rent ratio – primarily shape housing starts at the national level.
However, in the South and West, personal income, real GDP and property tax rates drive housing starts. The higher elasticities of housing starts to property tax rates at the regional level suggest that policymakers should exercise more caution in raising property taxes because they may drastically reduce the number of housing starts and residential investment, particularly in the Northeast.
This would cast doubt on the claim that Fannie Mae and Freddie Mac actually increase homeownership at the national and regional levels. Assuming the housing and finance sectors someday return to normal, and that this evidence from the past sheds light on the future, these results indicate that phasing out the GSEs would minimally affect housing starts and homeownership at the national and regional levels – at most, a 0.01% drop in the housing starts in the South and West and 0.005% drop in the Northeast and Midwest would occur.
Indeed, regardless of the level of aggregation or the region, it appears that economic fundamentals – and not conventional mortgage interest rates – primarily shape housing starts. The government and GSEs are leveraging the wrong instrument because subsidized mortgage interest rates have minimal impact on housing starts at the national and regional levels.
Moreover, land-use regulations play an important role in home price volatility and housing starts because home builders' beliefs are shaped by the regulatory environment and their investment returns, rather than by mortgage interest rates. Thus, it would seem that continuing to subsidize the mortgage interest rates is a waste of taxpayers' money because mortgage interest rates do not significantly influence home builders' decisions.
Nahid Anaraki is a visiting fellow for special projects in the Center for Data Analysis at the Heritage Foundation in Washington, D.C. She can be reached at (202) 546-4400. This article is adapted and edited from research published earlier this year by the Heritage Foundation.