RDP 2018-03: The Effect of Zoning on Housing Prices 8. Apartment Prices and Marginal Costs

This section compares apartment prices to the marginal cost of supply.^[22] The RBA's liaison with industry contacts has highlighted poor availability of suitable development sites (which is likely in large part due to zoning restrictions that do not allow high-density development in many areas) and the time and complexity of obtaining planning approvals as the most significant supply-side factors for new apartment construction in Australia (Shoory 2016).

As noted earlier, Glaeser et al (2005) argue that in Manhattan, the marginal cost of supplying apartments is equal to the construction costs associated with adding an additional storey to a new development. Although land is required to build an apartment block, this is a fixed cost – there is no marginal land cost involved with building higher on a given block. Glaeser et al then infer the presence of zoning restrictions by comparing apartment prices to these marginal construction costs.

Although Australia's large cities have ample space in urban areas that could be converted from other uses to apartment buildings, for simplicity we follow Glaeser et al by focusing only on the cost of building additional storeys.^[23] We rely on estimates of construction costs per square metre for apartment buildings of different heights provided by Rider Levett Bucknall (2017) and the CoreLogic unit record dataset to calculate average sale prices. A detailed discussion of our data and methodology is in the Online Appendix (Section 4).

For this exercise, we consider the marginal cost of constructing a 20-storey apartment building relative to a 10-storey building. RBA liaison evidence and Daley et al (2018, p 50) indicate that buildings in this range are representative of a large share of the apartment market in major cities. We make an allowance for other marginal costs associated with increasing the size of an apartment project, such as providing common areas, professional fees and an entrepreneurial return. Finally, we adjust costs for earlier time periods using the corresponding state PPI – output of other residential building construction series, and scale to the average size of apartments sold in each city during the corresponding period (e.g. 79 square metres in Sydney in 2016).

Our analysis of the Rider Levett Bucknall estimates suggests that, on average, it would cost about $470,000 to add an extra apartment to a 10–20 storey building in Sydney in 2016. This is the marginal cost; the average variable cost is $435,000. Although comparison can be difficult, alternative data sources point to similar, or somewhat smaller, estimates. For example, estimates by the CIE (2011) suggest a typical 100 square metre Sydney apartment costs $438,000 to build (adjusted to 2016 dollars using the appropriate PPI), whereas our estimates imply that such an apartment would cost $552,000 (average cost in a 10–20 storey building). Discussions with industry participants tend to suggest average construction costs cluster around $300,000 to $500,000, with more dispersion in estimates of fees, margins and overheads.

Our analysis of the CoreLogic dataset indicates that the average Sydney apartment sold for $870,000 in 2016.^[24] This is about $400,000 (85 per cent) more than our estimate of marginal cost, representing 46 per cent of the sale price. As with detached houses, this $400,000 wedge represents an arbitrage opportunity. The reason that builders do not exploit it is that they are prohibited from doing so by zoning regulations.^[25] See Moore (2012) for an example. Similar analysis, summarised in Table 7, implies that zoning restrictions added $120,000 to the cost of apartments in Melbourne and $110,000 in Brisbane. Our estimate of the zoning contribution for Sydney is broadly in line with the findings of Glaeser et al (2005), who found that zoning accounted for around half of the cost of apartments in Manhattan in 2002, but smaller than Lees (2018), who found that zoning accounted for around 70 per cent of the cost of apartments in Auckland.

If scarcity of government permission to develop apartments in desirable locations is pushing up their price, then we would expect the value or ‘shadow price’ of development rights to be capitalised into the prices of sites with such permissions. Discussions with developers and several examples in Appendix A are consistent with this. Site prices are often quoted as a price ‘per unit’, and our sense is that our estimated wedges from zoning of $400,000 per unit in Sydney and $120,000 in Melbourne are broadly in line with these prices.^[26]

Figure 7 plots average apartment sale prices against our estimates of marginal costs for Sydney, Melbourne and Brisbane over time.

Over the period from 1999 to 2006, we estimate that the gap between sale prices and marginal costs was low and broadly stable in all three cities, although the gap was slightly larger on average in Sydney ($121,000 versus $60,000 in Brisbane and $47,000 in Melbourne). This analysis suggests that zoning was not an important constraint on apartment development in Brisbane or Melbourne over this period.

In both Melbourne and Brisbane, marginal costs were broadly unchanged over the period from 2006 to 2016. However, average sale prices rose slightly, and so the gap between sale prices and costs has been somewhat higher in recent years. In Sydney, costs rose slightly over the same period, but apartment sales prices increased much more sharply. As a result, the gap between prices and costs – the zoning effect – has increased substantially in Sydney over the past decade.

Our estimates of apartment costs involve several assumptions and are somewhat uncertain. They are not comprehensive, but we believe they are representative of a large share of the apartment market. Moreover, the RBA's discussions with developers suggest that our overall numbers are in line with widely used rules of thumb in the industry. The results do not seem to be sensitive to moderate variations in the assumptions. More direct data would be useful. However, we would be surprised if it were to indicate a very different contribution of zoning to apartment prices.

Footnotes

We do not provide estimates for Perth owing to the low share of apartments in its dwelling stock. [22]

The lowest marginal cost way of supplying apartments could involve purchasing more land and building up fewer storeys, because the marginal cost of constructing apartments is an increasing function of building height due to the need to employ more expensive construction methods. This could mean that our estimates of the cost of supplying apartments are overstated. [23]

New apartments sold for somewhat more. Because we are using construction costs, we are likely to be overstating replacement value for apartments due to depreciation (renovations are less likely to provide a positive offset than in the case of detached dwellings). [24]

To the extent that planning regulations increase construction costs (rather than creating a wedge between costs and prices), for instance through prescriptive building design and materials requirements, then this effect of regulation on prices will not be captured in our estimates. [25]

The fact that site prices are commonly quoted in per unit terms, rather than in terms of the land area of the site, itself seems to be an indication that a physical scarcity of land is not the main factor driving high site prices. [26]