(This essay was published in Hong Kong Economic Journal on 30 October 2013)
The Long Term Housing Strategy (LTHS) consultation document issued by the Government-appointed Steering Committee in September 2013 is disappointing. It lacks ambitious goals and bold ideas. While it covers many fine points and suggests some reshuffling of priorities regarding the housing needs of the elderly, non-elderly singletons, inadequately housed households, youngsters, first-time buyers, and plot ratios, it offers few breakthrough ideas. The development of new towns, renewal of old urban areas, reforming of the small house policy, and tradeoffs regarding a more spacious living environment are all delegated as matters to be addressed beyond the next ten years. Older ideas like re-issuing Letters B to facilitate the redevelopment of agricultural land and reactivating the Tenant Purchase Scheme are not mentioned. There is also no mention of homeownership goals which have been a centerpiece in the housing policies of previous administrations.
The document concentrates on setting housing targets for the next 10 years. The heaviest criticism received to date has therefore been the timidity of its very modest housing targets. Over the next decade it aims to build only 470,000 units, 60% in the public sector and 40% in the private sector. It also leaves unanswered how some of the required land is to be found.
A target of building 47,000 units a year is arrived at on the basis of a projection that first puts the future annual increase in domestic housing demand at 29,400 new households each year. To this is added 17,600 new units to make up for a shortfall due to the demolition of old units, and to meet the needs of inadequately housed households, satisfying demand from external sources, and so on.
These unit targets, which are very modest, are based on a significant underestimation of the projected increase in long-term housing demand. The purpose of this essay is to explain the reasons for the underestimation. Both methodological problems and substantive issues are present. The implications for devising a sound long-term housing strategy given the substantive issues involved are then explored.
The core methodology driving the government’s projection of long-term housing demand is a two-step “life cycle” model. This approach is used in many countries. In the first step, future changes in the size of the population and its age-sex distribution are predicted. In the second step, the results obtained from the first step are used to predict future changes in the number of households (or headships). Past trends are assumed to be sustained into the future, i.e., 10 years in this instance. The long-term projected change in the number of households (or headships) would then constitute the projected change in the demand for housing.
Using this methodology, the annual net increase from 2011 to 2021 is projected to be 294,000 new households, which is slightly less than the actual increase of 315,000 households from 2001 to 2011. However, the figure for the latter decade does not reflect the fact that some households wanted to set up new households but could not afford to do so. In other words, the actual number of new households in 2001-2011, observed to be 315,000 households, should be seen as an understatement of the true underlying increase in housing demand. Therefore, there was unmet demand, which was why housing prices rose rapidly and sub-divided units appeared. The government’s approach underestimates this demand for housing and the growth of new households.
Theoretical Errors in the Projection Methodology
Let us examine theoretically the changes in the demand and supply of housing for the years 2001, 2011 and 2021 using Figure 1. In this diagram, the demand in year 2001 is denoted by the demand curve D2001, in year 2011 it increases to D2011, and in year 2021 it further increases to D2021. If over these same years the supply curve also increases by the same amount as the demand curves, then the increases in the number of households will be matched by increases in the number of housing units and these will be Q2001, Q2011, and Q2021.
However, if the increase in the supply of new housing units is less than the increase in the new household demand, a situation will emerge that is shown in Figure 2. The outcome will be a provision of Q*2001, Q*2011, and Q*2021 housing units in equilibrium. These numbers would underestimate the underlying growth in housing demand. In this second diagram, supply increases less than demand. How do households get accommodated under these circumstances? Housing prices will naturally have to rise. Some households will then be accommodated in the new housing units, some in sub-divided housing units, and some have to share their living space with other family members in older households even though they prefer to be separately accommodated.
Figure 2 is a better description of what actually happened in Hong Kong over the period 2001-2011. A situation that I believe is continuing in the decade up to 2021. The projections of the two step “life-cycle” model did not predict the hypothetical situation shown by Figure 1. Instead the model predicted the actual market outcome depicted in Figure 2. So instead of getting an increase in housing demand indicated by the numbers Q2001, Q2011, and Q2021 in Figure 1, it ended up predicting an increase in the number of households that were successfully accommodated indicated by Q*2001, Q*2011, and Q*2021 in Figure 2. The predictions assume therefore that increases in housing demand are the same as the number that can be accommodated in a tight housing market. But the number accommodated is a market outcome and will underestimate the underlying increase in market demand.
What is Wrong with the Projections?
What drives the increase in the demand for housing? Figure 3 shows that the percentage of single person households rose from 14.9% in 1996 to 15.9% in 2001, 16.5% in 2006, and 17.1% in 2011. The number of single person households over 5-year periods was 42,983 in 1991-96, 44,205 in 1996-2001, 46,542 in 2001-06, and 36,535 in 2006-11. This means about 7,000-8,000 new households were added each year. But these numbers are only market outcomes and therefore underestimate the underlying interest in forming new households because the wishes of some could not be accommodated for reasons of affordability.
Another driver of housing demand is the rapid acceleration of divorce and marriage rates (see Figure 4). The divorce rate has been rising upwards since 2001 and is still trending in that direction. In 1991, there were 31 divorced individuals households? per thousand households; the numbers increased to 70 in 2001 and 107 in 2011. In 2012, there were about 21,125 divorces, 40,841 first marriages, and 19,542 remarriages. At the present time, divorces and first marriages alone could increase new households by more than 60,000 per year and thus fuel demand for 60,000 more housing units, which is substantially above the 29,400 per year figure projected by the “life-cycle” model in the consultation document.
The “life-cycle” model is not a bad model to use for projecting future household demand for housing, but it inevitably produces underestimates when the future is projected on the basis of an extrapolation of past trends in a tight housing market. When housing demand is growing faster than housing supply, the number of households that can be accommodated will necessarily be less than the underlying increase in demand. By extrapolating past market outcomes into the future we will only be meeting part of the future increase in housing demand. This means the pressures that cause housing prices to rise and housing units to be sub-divided will continue.
Rising Divorce and Remarriage Rates and Housing Demand
We know in Hong Kong, as well as elsewhere, that divorce rates are higher among lower income households. This is indirectly evident from the higher proportion of divorced individuals living in rental units compared to homeownership units (see Table 1). Remarriage rates have increased significantly over time. Remarriage opportunities are much higher for men than they are for women, which means a growing proportion of divorced men are remarried, while women remain divorced. In 1991, within the whole population 21,140 men were divorced compared to 28,540 divorced women; the corresponding figures for 2001 are 55,620 and 92,420; and for 2011 they are 91,540 and 175,940.
Table 1: Housing Tenure of Married and Divorced Individuals, by Sex (numbers in thousands)
|Marital Status and Sex||Year||Public Renter||Private Renter||HOS||Private Owner||Total|
Note: Parenthesized values are percentages.
Source: HK Population Census and By-census, sample datasets.
What is apparent in Table 1 is that the number of divorced women living in public rental housing has increased rapidly compared to divorced men. In 1991, there were 8,260 divorced men and 9,220 divorced women living in public rental housing. By 2011 the figures had risen to 41,080 divorced men and 78,360 divorced women. It is likely that divorced women remain as public rental housing tenants while divorced men move out. Some of these divorced men who remarry subsequently would apply for public rental housing again if their incomes still qualified. The rise in applications on the waiting list for public rental housing units in recent years may be reflecting in part such a phenomenon. To what extent this is the case has to be further ascertained.
The current public rental housing allocation criteria favors married couples, but does not discriminate between first marriages or remarriages. Implicit in such an allocation criteria is a positive incentive that subsidizes divorces. At the same time, it also generates a penalty on children who inevitably suffer from broken families. The growing numbers of divorced women living in public rental housing units imply a rising number of children growing up in broken families in public rental housing estates. This is not conducive to upward social mobility, but sets the stage for the production of a new underclass that perpetuates intergenerational inequality.
Another likely consequence of rising divorce rates is that low income divorced men who move out of the public rental housing unit probably occupy a significant proportion of the sub-divided housing units. Some may remarry and eventually re-enter the public rental housing program, others may become permanently stranded in these private sector units.
Neither conjecture can be demonstrated quantitatively at this time as the data is not available. But I believe these conjectures should not be dismissed for they have grave implications for what kind of long-term housing strategy Hong Kong should adopt. As divorce and remarriage become increasingly prevalent in Hong Kong, the aims and consequences of our long-term housing strategies need a fundamental re-thinking. Are our present housing policies creating an underclass and making divorces too easy? The rise of sub-divided housing units is a symptom of a tight housing market, but it is also the product of the rising interest in forming new and separate households, some of which stem from rising divorce rates!
A housing strategy is not merely about how many housing units to build. It is also about what kind of society we will be encouraging through policies that influence household allocation choices. It is important that we think through more clearly and examine more carefully what is at stake. Are we encouraging divorces? Will we help create an underclass? Will we exacerbate future intergenerational inequality? Errors in forecasting demand can be easily remedied. Errors in policy will haunt many future generations.