House prices forecasted to drop 33,000 euros from peak, but expected to stabilize in 2024

The first months of 2023 are easy to summarize as far as the owner-occupied housing market is concerned: down further. Due to the aftermath of rising mortgage rates and further diminished confidence in the housing market, 11% fewer houses were sold in the first four months of the year than in the same period a year earlier. In April, house prices fell to 6.1% below their recent peak in the summer of 2022.

Both sales and prices have fallen slightly faster than we expected three months ago. Looking ahead, our forecast is therefore somewhat more conservative than in our March Quarterly Housing Market Report: We expect 165,000 sales in both 2023 and 2024 (down from 183,000 and 194,000, respectively). House prices are expected to be 5.2% lower on average this year than in 2022. Next year, as we now forecast, existing owner-occupied homes will cost on average 2.4% less than in 2023. In the last quarter of 2024, house prices will be 9.4% below the price peak in the summer of 2022, which is about EUR 33,000 lower for an average house sold.

The year-on-year change in house prices, which was up 21.1% early last year, fell to -4.4% in April this year (see Figure 1). Average homes were sold for an estimated EUR 19,000 less in April than a year earlier (see Figure 2). The average house price in April fell even more sharply by EUR 28,000. This is not only because homebuyers are paying less for the same home, but also because the market share of the cheaper segment has grown, so that the mix of houses sold has a lower price (a composition effect). Compared to the peak of the market, the average home sold is now EUR 26,000 cheaper.

The rest of 2023 will also be characterized by a rapidly cooling housing market. We expect house prices to be on average 5.2% lower this year than in 2022. Comparing the fourth quarter of this year with the fourth quarter of 2022, Statistics Netherlands’ (CBS) house price index is expected to be 6.5% lower (see Figure 3). Compared to the market’s peak in July 2022, owner-occupied homes will then be as much as 9.0% cheaper.

We do not expect house prices to fall much further next year. In the last quarter of 2024, we expect house prices to be only 0.4% lower than in the last quarter of 2023. However, due to the so-called carry-over effect, the average price level next year will still be 2.4% lower than in 2023. The year-on-year decline in the average price level is almost entirely the result of price declines that we expect to occur before 2024 (see Figure 3).

The expectations that house prices will stabilize next year are related to the still tight housing market and the quickly improving affordability of existing owner-occupied homes this year. Interest rates will also play a role next year. Capital market interest rates, which largely determine the level of mortgage interest rates, are expected to peak this year, only to decline somewhat thereafter. Moreover, a recession has not yet occurred, and slight GDP growth of 0.6% and 0.9% is expected for 2023 and 2024, respectively. The unemployment rate, while gradually rising from 3.5% in 2022 to 4.0% in 2024, according to our estimates, remains very low by historical standards.

Since our last quarterly report, we have revised our estimate of house prices slightly downward. At that time, we assumed a decrease of 4.2% this year and 1.5% next year. House prices are currently falling at a slightly faster pace than we assumed last quarter, and the economic picture has also become (slightly) more negative. However, the overall picture – that the housing market will stabilize after this year – remains unchanged.

Maximum mortgage in 2024 back to early 2022 levels

The affordability of owner-occupied homes has improved this year as house prices have fallen. Substantial wage increases also contribute to this. The Netherlands Bureau for Economic Policy Analysis (CPB) expects gross typically earned incomes to be 6.5% higher this year than in 2022, and a further 4.9% higher in 2024. With such increases, potential homebuyers are expected to be able to borrow almost as much in 2024 as they could in early 2022, when mortgage rates were about 3%-points lower (see Figure 4). We expect capital market interest rates to be 0.4%-points lower on average next year than they were this year. Should mortgage rates also fall by 0.4%-points, potential homebuyers could even borrow a fraction more next year than they did in early 2022. Combined with lower house prices and a larger number of homes for sale, homebuyers should have more options in the housing market in 2024 than they did in mid-2022.

Housing market outlook depends on development of capital market interest rates

In our baseline scenario, we assume that capital market interest rates will reach their provisional peak this year. However, future interest rate developments are highly uncertain. Therefore, as in previous quarterly reports, we have calculated two alternative scenarios: a scenario with a 1%-point lower interest rate and a scenario with a 1%-point a higher interest rate. In the baseline scenario, the ten-year swap rate averages 2.9% this year, to reach 2.5% next year. In the alternative scenarios, the interest rate differential of 1%-point will be reached in the first quarter of 2024, after a gradual rise in the last two quarters of 2023. In the lower-interest rate scenario, the swap rate will only be 1.5% in 2024, while in the high-interest rate scenario it will rise further to 3.5% that year (see Figure 5).

The gradual increase in the interest rate differential relative to the baseline scenario – combined with the delayed impact of interest rate developments on house prices – means that in 2023 there is virtually no difference between house prices in the different scenarios. As from 2024, however, clear differences can be seen. While in the baseline scenario house prices will practically stabilize after 2023, in the high-interest rate scenario house prices will fall even further. The year-on-year decline in house prices in this scenario will reach 3.3% in 2024. In the low-interest rate scenario, house prices will start to rise again in 2024. Yet, due to the spillover effect, even in this scenario there is still a 1.5% year-on-year decline in house prices in 2024. By the last quarter of 2024, however, house prices will already be 1.0% higher than in the same quarter of 2023.

Amsterdam and Utrecht have seen the sharpest drops in house prices since their peak. Prices fell 7.9% and 10.2%, respectively (see Figure 4). Of the four major cities, these are also the two where prices rose the most since their previous price peak: Between 2008 and the peak of the market in mid-2022, owner-occupied homes in Amsterdam rose 97% in value. In Utrecht, prices went up as much as 101% over the same period (compared with a national increase of 58% on average). Of all provinces, prices in Flevoland went up the most since 2008, at 78%, and after the provinces of Utrecht and Noord-Holland, this is where prices have now fallen the most so far.

Regional house price trends

House prices are expected to fall in all regions this year, but there are significant differences in the pace (see Figure 7). While house prices in Zeeland, parts of Overijssel, and northern Limburg are expected to fall by a modest 3%, the price drop in the Amsterdam and Utrecht area is forecasted to be as much as 8%. In the Groningen area, which includes the city of Groningen, we also foresee a sharp drop in prices.

There are several explanations for these divergent growth rates. The cooling of the housing market is mainly caused by the fact that rising interest rates have worsened the affordability of owner-occupied homes. This may have a greater impact in regions where affordability has been under severe pressure for some time – such as Amsterdam and Utrecht – than in regions with lower prices. For example, in recent years it has contributed to increased migration from the urban areas of the Randstad to cheaper areas outside the Randstad. The rental market can also play a role. In places such as Amsterdam and Groningen, the role of real estate investors has been relatively large, but demand for rental housing has declined due to a combination of factors. These include lower initial returns, higher property transfer taxes on investment homes, higher taxation of real estate investments in box 3, and the announced regulation of rents in a large part of the private rental segment.

The number of transactions has fallen considerably in recent years (see Figure 8). In the first four months of 2023, only 53,000 existing owner-occupied homes changed owners – 11% fewer than in the same period a year earlier. There has not been a spring with fewer home sales since 2015. Last year, due to the lack of supply, there were already fewer transactions than in 2020 and 2021, when 70,000 and – thanks to amendments to the property transfer tax – 86,000 homes were sold in the first four months, respectively.

Despite the cooling housing market, the rate of transaction decline has actually slowed since mid-2022. This is probably because there are still many people who would like to buy a home, but have had to postpone their plans in recent years due to lack of supply. This includes first-time buyers who kept getting outbid.

With more houses for sale and prices dropping, new opportunities are opening up for some potential homebuyers. Existing homeowners can also transfer their current mortgage to their new home under certain circumstances and conditions, sometimes still benefiting from a relatively low interest rate. Also, many homeowners still have a considerable surplus value on their property. As a result, they are more likely to accept a lower offer for their current home, providing prospects for potential homebuyers.

The number of sales in the first quarter of this year was lower throughout the Netherlands than in the same quarter last year, with the exception of the province of Utrecht (see Figure 9). On average, the decrease was 7.9%, but in Friesland the number of transfers fell by as much as 18.6%. The decline was also relatively strong in Drenthe, Flevoland, Limburg, and Zeeland, where it was almost 15%. Looking at the major cities, it is striking that the number of transactions fell the least in the regions where prices fell the most – Utrecht and Amsterdam. It is unclear why.

Over-35s throw in the towel

Both young and old buyers bought fewer existing homes in the first months of the year, but the drop is steepest among those over 35. The number of sales to 35- to 54-year-olds in the first four months of this year was 19% lower than a year earlier, and those over 55 bought almost 14% fewer houses. Among the Dutch under 35, the decrease was only 4% (see Figure 10). This seems counterintuitive from a financial point of view, since it is potential first-time buyers (more likely to be under 35) who are most dependent on mortgage financing and thus feel the increased interest rates most strongly in their home purchasing power and monthly costs. Yet, this picture is similar to that of 15 years ago, when young adults also dropped out to a lesser extent than those over 35 in the aftermath of the financial crisis (see Figure 11).

Residential relocations by 35-year-olds are more likely to be driven by life events – such as moving out of one’s parents’ home, moving in with a partner, or starting a job. These motives are likely to remain strong, regardless of developments in the housing market. Life events also play a role in the propensity to move among people in the Netherlands over the age of 35, but those over 35 are usually less on tenterhooks. After all, they often already live independently, usually in an owner-occupied home. So for them, moving is usually not a matter of buying a house, but of buying another house. That is a desire that’s easier to give up when the housing market cools and homeowners doubt the marketability of their current home. Research by the Dutch Land Registry also shows that housing investors (more often over 35) have been buying fewer houses for a number of years now, which may be an additional explanation for the relatively large drop in the number of home purchases by people over 35.

The fact that the number of sales among young adults has fallen less sharply – and thus their ‘market share’ in the owner-occupied housing market is currently relatively high – does not automatically mean that homeownership among those in their twenties and thirties is increasing. In fact, relative to the number of under-35s, fewer houses are being sold now than in recent years, when historically relatively few young adults bought a house (see Figure 12) and homeownership fell sharply (see Figure 13 and our earlier publication “This is how inaccessible owner-occupied housing has become for first-time buyers,” in Dutch only).

We expect the number of existing home transactions this year to be considerably lower than last year, at 165,000 sales. For next year, too, we foresee 165,000 sales. This is a substantial downward revision from our previous quarterly report, when we still assumed 183,000 transactions this year and 193,000 next year. The downward revision is mainly due to new insights into the relationship between new-build home sales and the number of existing home transactions.

As of this round of estimates we are using a newly developed forecasting model, in addition to the usual tools (see Appendix). In this model, the number of new-build owner-occupied home sales turns out to play a major role in explaining transactions of existing owner-occupied homes. As we are currently seeing sales of new-build homes plummet – as much as 50% fewer new-build homes were sold in the first quarter of this year than in the same quarter last year – there will be fewer homeowners moving in the market for existing owner-occupied homes in the coming years. While we expect the marketability of existing owner-occupied homes to gradually improve, significantly fewer homes will be put up for sale.

The new model can also be used to calculate the effect of interest rate scenarios on the number of transactions. As with house prices, we look at a scenario with a 1%-point lower interest rate and a scenario with a 1%-point higher interest rate. In the lower-interest-rate scenario, the number of transactions will increase considerably next year. In this scenario, we expect 167,000 transactions this year and 177,000 next year. In the higher-interest-rate scenario, the number of transactions will be hit hard again next year, resulting in 163,000 transactions this year and 155,000 next year.

To predict the number of transactions, we have developed a new econometric model that predicts both the number of new-build home sales and the number of sales of existing owner-occupied homes. The number of new-build home sales plays an important role for the market for existing owner-occupied homes because the sale of new-build homes triggers a ‘chain’ of residential movement. New-build homes are often bought by people who already own a home. Some of the buyers of the homes they leave behind already own a home themselves, which then comes back on the market and, in turn, leads to additional homeowners moving. The chain ends when a first-time buyer purchases a vacant owner-occupied home.

We base our modeling of the number of new-build home sales on the year-on-year change in the existing owner-occupied homes price index, long-term interest rates, unemployment, an economic variable (based on the deviation of gross domestic product from its long-term trend), a linear trend, and quarterly dummies. Figure 15 shows both the actual number of new homes sold and the forecast produced by the model. The R² for this model is 0.70. That means the model can explain 70% of the variation in the number of new homes sold. The model predicts that 58% fewer new-build homes will be sold this year than in 2021, and 51% fewer next year than in 2021.

We base our modeling of the number of transactions of existing owner-occupied homes on the four-quarter moving average of the number of new-build home sales, the year-on-year change in long-term interest rates, a linear trend, and quarterly dummies. Figure 16 plots the actual number of transactions against the number of transactions predicted by the model. The R² in this case is 0.80. Of all the variables, new-build home sales have the highest explanatory power in this model.

For both models, we tested whether they meet the stationarity and/or cointegration requirements for time series analysis. For verification, we have also estimated a model in which the number of transactions is explained all at once (without an intermediate step in which the number of new homes sold is predicted first). This model produces similar predictions for the coming years, but performs significantly worse, especially in the years before and after the 2008 financial crisis. At first there was a high level of construction, but then new construction collapsed. Taking this into account provides a better explanation for the development of the number of transactions.