Impacts on the House Price Distribution

Mean house price changes may fail to capture the effect of population changes if changes in housing demand are focused in particular parts of the house price distribution. For example, Returning Kiwis have relatively high average incomes, suggesting that they may have a grater influence on demand for higher-price housing. They are however, like New Immigrants, relatively young and may therefore exert more pressure on the market for lower priced first-homes.

Table 9 shows the relationship between the components of population change for each area and different quantiles of the local house price distribution. Inflows of Returning Kiwis are most strongly related to house price increases at the 25th percentile of the house price distribution (elasticity of 10.4), although around half of this effect is accounted for by observable changes in local population characteristics, some of which may be a result of the different characteristics of the Returning Kiwis. The changes in median house prices and upper quartile house prices are somewhat lower than at the lower quartile, although still high, with elasticities of 7 to 8. The patterns are consistent across different geographic area definitions, although the elasticities are estimated with less precision. Overall, the general pattern of effects estimated for mean house prices are also evident at other points in the local house price distribution, suggesting that population changes have similar impacts in relative terms on both cheaper and more expensive homes in local areas experiencing these changes.

Sub-Period Differences

The descriptive results in Figure 3 indicate that the relationship between population changes and house price changes differs across the four intercensal periods. In this section, we re-estimate equations (2) and (4) allowing the overall population elasticity in equation (2) and the elasticities of each population component in equation () to vary by period. This extension allows us to test whether our regression results are robust across time-periods or follow a similar pattern as the descriptive results.

The resulting estimates are presented in Table 10. The first panel contains population change elasticities for each period and shows that the finding of different elasticities in different periods still holds when we control for local population characteristics and when we examine other housing price measures.^[20] The remaining panels show the estimates of the period-by-period elasticities for each population component, with each column containing estimates from a single regression. These results reveal a number of variations from our general findings. Inflows of Returning Kiwis are not significantly associated with house prices in 1986-1991, and are associated with relatively less house price appreciation in 2001-2006, compared with 1991-1996 and 1996-2001. New Immigrant inflows are associated with higher house prices in both 1986-1991 (without population covariates) and 1991-1996, and strongly declining house prices in 1996-2001 (marginally significant) and 2001-2006 (without population covariates). In both 1991-1996 and 2001-2006, changes in the number of Local Kiwis and Previous Immigrants are more strongly associated with house prices than in other periods, with areas experiencing increases in the number of Local Kiwis having lower house price appreciation and areas with increased numbers of Previous Immigrants having higher appreciation.

The lack of consistency in these estimates across the different periods suggests that the relationship between components of population change and house price movements is more complex than can be captured by our estimation. At the least, it suggests that there are omitted or mediating factors that may be more important than population change per se in determining house prices. Further consideration is warranted of cyclical influences at the aggregate or LMA level. Unfortunately, the five-yearly frequency of our data means it not ideally suited for analysis along this dimension.

Neighbourhood Housing Markets

One potential criticism of our estimation approach is that immigrants may have a strong effect on neighbourhood housing markets without affecting the more aggregated areas on which our analyses so far focus. To assess the strength of the relationship between population changes and neighbourhood housing prices, we re-estimate equation (4), but at the level of census areas units and include fixed effects for each LMA-year combination.^[21] Area units are aggregations of census meshblocks, typically contain around 2,000 people, and correspond to suburbs within cities, and to somewhat larger areas outside cities. Because we include fixed effects for each LMA-year combination, our estimates are based on variation in changes in the composition of neighbourhood populations within local labour market areas, and control for the fact that different areas in New Zealand are more or less attractive to New Immigrants and Returning Kiwis and have higher or lower house prices.

The resulting estimates are presented in Table 11 and show a similar pattern to those for more aggregated geographical areas (Table 6), although the magnitudes of the effects are smaller. In neighbourhoods where Returning Kiwis add an additional one percent to the population, house prices rise by 1.3 percent, controlling for changes in neighbourhood population characteristics. The rise is somewhat smaller and statistically insignificant for flat sales prices (0.8 percent), for housing rentals (0.2 percent), and for flat rentals (0.2 percent). These results indicate that the relationship between components of population change and changes in house prices is evident both in aggregated housing markets, whether these are defined as local labour market areas, TLAs or RCs, and in local neighbourhoods, controlling for the greater area effects.

To summarise, neighbourhoods experiencing relatively higher population growth from Returning Kiwis have greater house price increases than the rest of their LMA regardless to the overall scale of the Returning Kiwi inflow to the LMA. Opposite results are found for inflows of New Immigrants, with neighbourhoods with relatively larger inflows experiencing slower house price growth relative to the LMA in general. Further investigation is warranted into whether the intra-LMA relationship between neighbourhood population changes and neighbourhood house price appreciation results from the sorting of different population groups into different neighbourhoods and/or from highly localised differential impacts on housing demand due to imperfect substitutability of housing in different neighbourhoods and neighbourhood variation in housing supply responsiveness.