Analysis No. 5 · November 2026 · Nelson Viegas
Portugal is not short of homes. It is short of homes where people want to live.
I gathered 25 years of data on prices, credit, construction, immigration, tourism and the Census, from Statistics Portugal (INE), Banco de Portugal, the BIS and academic studies, and crossed them region by region. No single factor explains the rise on its own. What explains why all of them passed into the price is the housing that stopped being built where demand is concentrated.
New homes for each new household in Grande Lisboa, Península de Setúbal, Grande Porto and the Algarve, between 2011 and 2020; in the rest of the country it was 3.19
INE, Official; calculationNew homes completed in Portugal in 2015; in 2002 there had been 125,708
INE, OfficialMunicipalities where the price per square metre more than doubled between 2017 and 2025
Banco de Portugal, OfficialOf average earnings to pay for 90 m² in the Algarve; nationally it is 8.5 years
INE and GEP, Official; calculation
The first four analyses looked at business-to-business sales and at consumer goods. This one changes market and goes to the home, the biggest purchase most families make in their lives.
It is a sector I know from the inside. I worked very actively in real estate, mainly in real-estate marketing and sales. Until 2013, Portuguese estate agents and property finders had to prove their professional capacity to renew the estate agency licence (AMI) and the property finder registration with the InCI, the sector regulator. There were two ways to do it: an exam or continuing-training credits. My training courses in real-estate marketing and sales gave those credits.
I wanted to understand why homes became so expensive, without starting from a ready-made answer. I let the data speak first, block by block, from 2000 to 2025, and the thesis appeared at the end. It is not the one you hear most often.
I wrote with families who want to buy or rent a home in mind. At the end, I bring together what all this means for them today.
As in the previous analyses, when the official sources disagree, I show the numbers side by side instead of picking one.
Nelson ViegasFounder of Superior Training Technology
Summary
In chapter order. Each one leads to the evidence.
Between 2001 and 2007 the real price fell 9%, while families took on debt and Spain almost doubled its own.
See the evidence →19.6 billion euros of new lending in 2007 and a first wave of immigration, with no real price rise. Construction had been falling since 2003.
See the evidence →New lending fell 90% and the real price ended 34% below its peak. At the same time came golden visas, the new rent law and the short-term rental regime.
See the evidence →In 2015, 7,148 new homes were completed, 5.7% of the 2002 peak. Prices had been rising since the end of 2013.
See the evidence →In Santa Maria Maior, 28% fewer family homes in a decade. Nationally, short-term rentals are about 2% of homes.
See the evidence →Non-residents accounted for 15.7% of the value of sales in 2014 and 8.2% in 2025, the year of the biggest price rise.
See the evidence →Immigration, interest rates, record lending and support for young buyers: in 2025 prices rose 17.6%, the highest in the series.
See the evidence →Lisbon, Setúbal, Porto and the Algarve had 74% of the decade's new households and 34% of its new homes. In the rest of the country, three homes were built for each new household.
See the evidence →How many homes are built, how many immigrants there are, how many homes are missing: the official sources disagree on eleven key quantities.
See the evidence →Credit, tourism, foreign buyers and immigration weighed at different moments. With the construction of the early 2000s, none of them would have had the same effect.
See the evidence →The spine of the analysis
The mechanism has three parts, and every chapter comes back to them. The first is new supply: it fell from 2003 and never recovered where demand is concentrated. The second is the waves of demand, which arrived one after another: tourists, investors from abroad, immigrants, young people with a public guarantee. The third is geography: each wave passes first into the price of the centre and then spreads outwards. When there was construction, the waves of demand did not pass into the price. Since it stopped, all of them have.
Portugal is not short of homes: it is short of homes where people want to live. It stopped building in those places from 2003 and, when demand came back in 2013, every wave that arrived (tourists, investors, immigrants, young people with a public guarantee) passed entirely into the price, first in the centre and then in waves outwards.
Each of these factors is real; none works without stalled supply. In this analysis, they come in as the waves that passed into the price.
Today's price rise began in the third quarter of 2013, but the forces behind it began at different dates, some much earlier. Construction began to fall in 2003; wage stagnation is documented from 2002; the first wave of immigration was in 2006; short-term rentals entered the law in 2008 and exploded in 2014. The chart shows the curve of real prices in Portugal and Spain; the table puts the starting dates side by side.
In real terms, Portugal fell 9% between 2001 and 2007 while Spain almost doubled. Since the 2013 trough, Portugal has risen 129% and is 53% above its 2001 peak. Spain has risen 49% from its trough and is still 15% below its 2007 peak1, 3calculation.
| Trend | Begins | Starting signal | Source |
|---|---|---|---|
| Stagnation of real wages | 2002 | Average real base wage +5.2% in 16 years, from 2002 to 2017 | 23 |
| Fall in new construction | 2003 | After the 2002 peak; in 2013 the index stood at 22.9 (2000 = 100) | 7 |
| Fall in real prices | 2001 | Real peak in Q4 2000 | 1 |
| Household debt | 1990s, accelerating 2000–2006 | From 86% to 124% of disposable income | 20 |
| First wave of immigration | 2006 | +57,000 foreign residents in one year, +65% from 2005 to 2009 | 5 |
| Peak of new lending | 2007 | 19.6 billion euros, highest in the series | 18 |
| Short-term rentals in the law | 2008 | Legal category created; almost no registrations until 2013 | Decree-Law 39/200835, 86 |
| Growing share of renovation | 2008 | From 20.4% of completed works to a third in 2015 | 7, 8 |
| Tax benefits for new residents | 2009 | Non-Habitual Resident regime | Decree-Law 249/200987 |
| Golden visas | October 2012 | 11,628 up to January 2023 | 26 |
| Liberalisation of rents and eviction | 2012–2013 | National Rental Office, about 3,000 evictions a year | 41 |
| Price rise | Q3 2013 | Real trough in Q2 2013 (80.0) | 1 |
| Short-term rental boom | 2014 | Own regime; 14,127 registrations at the end of 2014, 77,053 in October 2018 | 30 |
| Recovery of lending | 2014 | 3.9 billion, almost double 2013 | 17 |
| Recovery of permits | 2015 | From 8,618 new homes permitted (2015) to 24,905 (2019) | 9 |
| Negative interest rates | 2016 | First negative annual average of 12-month Euribor | 4 |
| Second wave of immigration | 2016, accelerating in 2018 | From 383,759 (2015) to 1,543,697 (2024) | 5, 6 |
| Rising interest rates | 2022 | From −0.49% (2021) to 3.87% (2023) | 4 |
| Foreign residents in mortgage lending | 2021 | From 7.38% to 13.56% of the amount (2021–2025) | 45 |
Two dates deserve attention. Construction began to fall ten years before prices began to rise: when demand came back in 2013, the capacity to build had already gone. And the price rise began before the short-term rental boom, negative interest rates and the second wave of immigration: these factors sped up a rise that was already under way.
Before you read
The same decade of prices can be told as a rise of 164%, 175%, 180% or almost 240%. All the sums can be right: they measure different things. Before reading the blocks, it helps to know which ruler is being used.
A note on the BIS series before 2009: the source itself changes method that year. The idea that Portugal had no bubble in the 2000s rests partly on valuations, not on sales. It is a weakness to keep in mind, not a reason to set the series aside: the 2014 Banco de Portugal study that compared Portugal and Spain also rested on bank valuations and reached the same conclusionsecondary.
| Who says it | How much it rose | Period and ruler |
|---|---|---|
| Euronews (June 2026), with Global Property Guide data27 | almost 240% | "last decade", ruler not stated |
| idealista, from INE | 180% | "last decade" to 2025secondary |
| BIS, Q4 2015 to Q4 20252 | 175% | nominal, quarter on quartercalculation |
| BIS, 2015 average to 2025 average2 | 164% | nominal, annual average on annual averagecalculation |
| BIS, 2015 average to 2025 average1 | 114% | real, adjusted for inflationcalculation |
The difference between 164% and 240% is not a rounding error. It comes from the starting point (the 2013 trough gives bigger numbers than 2015), from the ruler (index or median) and from using annual averages or quarters. In this analysis, every figure for a rise says where it starts and which ruler it uses.
Between 2000 and 2025 the nominal index rose 160%, from 94.4 to 245.8 (base 2010 = 100). The real index rose 55%, from 120.1 to 186.11, 2calculation. The difference is accumulated inflation. At two moments the distinction changes the story:
In this analysis, the blocks use the real index to compare periods and the nominal one to talk about what people paid.
To read it better
The technical terms that appear in the analysis, explained in a few words.
Chapter 1 · 2000–2004
At the start of the euro, Portugal had falling interest rates, easy credit and the biggest construction machine in its recent history. Even so, the real price of homes fell 6% between 2001 and 20041calculation. This is the block that contradicts the idea that low interest rates always push prices up.
| Variable | 2000 | 2001 | 2002 | 2003 | 2004 | Source |
|---|---|---|---|---|---|---|
| Nominal index (2010 = 100) | 94.4 | 99.4 | 100.1 | 101.2 | 101.9 | 2calculation |
| Real index (2010 = 100) | 120.1 | 121.3 | 117.9 | 115.4 | 113.5 | 1calculation |
| Real change | +4.8% | +1.0% | −2.8% | −2.1% | −1.6% | 1calculation |
| 12-month Euribor, annual average | 4.78% | 4.08% | 3.49% | 2.34% | 2.27% | 4calculation |
| Foreign population with a residence permit | 207,587 | 223,997 | 238,929 | 249,995 | 263,322 | 5 |
| Monthly minimum wage (€) | 318.2 | 334.2 | 348.0 | 356.6 | 365.6 | 22 |
Between 2000 and 2004, an average of about 104,000 new homes were completed each year14. The peak was in 2002: 125,708 homes completed in new construction for family housing, according to the INE series in PORDATAsecondary. INE itself, in 2014, put that year at 113.4 on an index where 2000 equals 1007.
The 2021 Census gives the measure of what remained: 529,510 of the existing buildings (14.8%) were built between 2001 and 2010. Those built between 2011 and 2021 number 110,784, only 3.1%12. One decade made five times more buildings than the next.
At the end of 2000, household debt was worth 86% of disposable income. At the end of 2006 it was worth 124%20. The 12-month Euribor fell from 4.78% in 2000 to 2.27% in 20044calculation, and banks lengthened loan terms between 1995 and 200620.
Subsidised credit, which subsidised interest on home purchases, ended for new contracts with Law 16-A/200225exact date of entry into force to be confirmed in the Diário da República. The price series shows no break linked to that end: the real fall had been under way since 2001.
Banco de Portugal wrote in 2007 that "there is no evidence of overvaluation of prices in the residential property market"20. The data for this block points to three reasons, which reinforce each other:
In the same block, the real price in Spain rose 55%, from the 2000 average to the 2004 average3calculation. The two countries had the same euro, the same Euribor and banks lending easily. The difference lay elsewhere. Spain had its construction peak in 2006, with 658,510 homes completed, at the end of the bubble; the Portuguese peak was in 2002, before real prices began to fall85.
Chapter 2 · 2005–2009
New mortgage lending reached the highest level in the series in 2007, at 19.6 billion euros18, 19. In the same year, the real price of homes fell 1.9%1calculation. This block also contains a wave of immigration that hardly appears in today's narratives.
| Variable | 2005 | 2006 | 2007 | 2008 | 2009 | Source |
|---|---|---|---|---|---|---|
| Nominal index (2010 = 100) | 104.2 | 106.3 | 106.9 | 100.2 | 99.2 | 2calculation |
| Real index (2010 = 100) | 113.5 | 112.4 | 110.3 | 100.7 | 100.6 | 1calculation |
| Real change | 0.0% | −1.0% | −1.9% | −8.7% | −0.1% | 1calculation |
| 12-month Euribor, annual average | 2.33% | 3.44% | 4.45% | 4.81% | 1.62% | 4calculation |
| Foreign population with a residence permit | 274,631 | 332,137 | 401,612 | 436,020 | 451,742 | 5 |
| Dwellings completed, all types of works | — | — | — | 68,336 | 55,097 | 7 |
| Dwellings permitted, all types of works | — | — | — | 55,606 | 35,585 | 7 |
| Monthly minimum wage (€) | 374.7 | 385.9 | 403.0 | 426.0 | 450.0 | 22 |
The 19.6 billion euros of 2007 are the highest in Banco de Portugal's series of new loans, which starts in 200318. Household debt was already at 124% of disposable income at the end of 200620. Measured as a percentage of GDP, household debt peaked in 2009, at 91%21.
The 12-month Euribor more than doubled between 2005 and 2008, from 2.33% to 4.81%4calculation. Since most loans were variable-rate (still 80.2% in November 202218), monthly payments rose without families having borrowed more. In 2009, with the crisis, Euribor fell to 1.62%.
Between 2005 and 2009, the foreign population with a residence permit rose from 274,631 to 451,742 people, 65% more in four years5calculation. From 2006 to 2007 alone, 69,000 more arrived. The peak of this wave, 451,742 in 2009, was only exceeded again in 20185.
In these years real prices fell. It is a fact that weighs in today's debate on immigration and housing: a 65% rise in the foreign population coincided with falling prices, at a time when tens of thousands of homes were being built each year.
A caveat: part of the 2006–2008 jump may come from legal changes in the way people were counted (the 2007 foreigners law changed the types of permit), and not only from new arrivals. It is flaggedto be confirmed in the SEF reports for 2006 to 2008.
In 2008, 68,336 dwellings of all types of works were completed, and in 2009, 55,097. Permits fell faster: from 55,606 to 35,5857. For 2007, Banco de Portugal gives 67,500 new homes completed16. For 2008, the INE series in PORDATA gives 59,256 dwellings completed in new construction for family housingsecondary. The two 2008 numbers do not contradict each other: the first includes extensions, alterations and reconstructions.
Construction halved in six years, and it fell before the crisis. When credit shut down in 2008, the machine was already slowing.
In 2008 the real price fell 8.7% in Portugal1calculation. In Spain, between the 2007 and 2009 averages, it fell about 11%3calculation. In this first year of crisis, the two countries fell in a similar way. The difference came afterwards: the Spanish fall continued until 2013 and was much deeper (next block).
In Ireland, according to Eurostat, prices fell 50% from the peak at the end of 200751.
Chapter 3 · 2010–2014
Between 2010 and 2013 the real price fell 19%, and the trough came in Q2 20131calculation. In the same block, almost all the rules that shaped the following decade were approved: golden visas, a new rent law, the urban renovation regime, a dedicated regime for short-term rentals. This is the block in which today's trends are born.
| Variable | 2010 | 2011 | 2012 | 2013 | 2014 | Source |
|---|---|---|---|---|---|---|
| Nominal index (2010 = 100) | 100.0 | 95.1 | 88.4 | 86.7 | 90.4 | 2calculation |
| Real index (2010 = 100) | 100.0 | 91.8 | 83.0 | 81.2 | 84.9 | 1calculation |
| Real change | −0.6% | −8.2% | −9.6% | −2.1% | +4.5% | 1calculation |
| 12-month Euribor, annual average | 1.35% | 2.01% | 1.11% | 0.54% | 0.48% | 4calculation |
| Mortgage lending granted (€ million) | 4,710 | 1,847 | 2,032 | 2,372 | 3,912 | 17calculation: monthly average × 12 |
| Home sales | 129,950 | −28.0% | −18.4% | 79,775 | — | 7, 8 |
| Dwellings completed, all types of works | 41,368 | 32,379 | 34,357 | 26,242 | — | 7 |
| Dwellings permitted, all types of works | 31,394 | 24,687 | 17,967 | 10,519 | — | 7 |
| Foreign population with a residence permit | 443,055 | 434,708 | 414,610 | 398,268 | 390,113 | 5 |
| Monthly minimum wage (€) | 475 | 485 | 485 | 485 | 485 | 22 |
Mortgage lending granted fell from 19.6 billion euros in 2007 to about 1.8 to 2.0 billion in 2011 and 201217, 19. That is a fall of about 90%. Bank valuations of homes, the first step in a mortgage, fell from about 125,000 in 2009 to about 31,600 in 20137.
Sales followed: 129,950 in 2010, the highest in the series until 2016, and 79,775 in 20137, 8. The share of existing homes in sales rose from 63.8% to 73.4%7. New homes almost stopped being sold, because they almost stopped being built.
The foreign population fell 14% between 2009 and 2014, from 451,742 to 390,1135calculation. Estimates of Portuguese emigration point to between 100,000 and 135,000 people a year and negative net migration in every year of the troikato be confirmed with INE. The minimum wage was frozen at 485 euros from 2011 to 201422, and the average real wage fell from 201123.
Credit shut down, incomes falling and people leaving: no factor was pulling prices up, and none did.
Dwellings permitted fell to a third in four years: from 31,394 in 2010 to 10,519 in 20137. In 2013 only 7,286 new homes for family housing were permitted7.
At the same time, the share of renovation in completed buildings rose: 20.4% in 2008, 25.6% in 2011, 26.8% in 2012 and 29.1% in 20137, 8. In number, renovation did not grow: 6,954 renovated buildings were completed in 2012 and 6,715 in 20137. The share rose because new construction fell more. It is the first clue to the phenomenon the chapter on short-term rentals develops.
| Year | Measure | What changed | Verification status |
|---|---|---|---|
| 2009 | Non-Habitual Resident regime | Tax benefit for people who become resident in Portugal | Diploma to be confirmed |
| 2012 | Revision of urban leases (Law 31/2012) | Updating of old rents and faster eviction, with the National Rental Office from 2013 | Law to be confirmed; evictions:41 |
| 2012 | Revision of the legal regime for urban renovation (Law 32/2012) | Simplifies renovation in urban renovation areas | Diploma located, content to be summarised |
| Oct. 2012 | Residence Permit for Investment (golden visas) | Residence in exchange for buying property from 500,000 euros, among other routes | Totals:26 |
| 2014 | Exceptional regime for urban renovation (DL 53/2014) | Exempts renovation works from some technical rules designed for new construction | Diploma located |
| 2014 | Dedicated short-term rental regime (DL 128/2014, in force in Q3) | Short-term rental stops being a residual category | Date:38 |
The National Rental Office recorded 38,834 evictions since 2013, a little over three thousand a year on average41. The Directorate-General for the Administration of Justice warns that the real number may be lower, because the handover of the property is not always reported41.
In Santa Maria Maior, the parish of Alfama, Baixa and Castelo, new short-term rental registrations were 21 in 2010, 9 in 2011, 42 in 2012, 43 in 2013 and 243 in 201435. Across the country, at the end of 2014, there were 14,127 registrations30. In Lisbon, Airbnb listings were about 2,000 in 201436. The jump begins exactly in the year of the new regime.
From the peak (Q3 2007) to the trough (Q4 2013), the real price in Spain fell 44.5%3calculation. In Portugal, from Q1 2007 to the trough in Q2 2013, it fell 28.7%1calculation. Spain fell more because it had risen more: the Spanish real price almost doubled between 2000 and 2007, and the Portuguese one fell.
Chapter 4 · 2015–2019
Between 2013 and 2019 the real price rose 47%, with almost zero inflation and negative Euribor1, 4calculation. New construction stayed at its lowest of the century. Short-term rental registrations went from 24,000 to more than 90,000, and non-resident buyers reached 13% of the value of sales. It is the block in which the most new factors come in at once.
| Variable | 2015 | 2016 | 2017 | 2018 | 2019 | Source |
|---|---|---|---|---|---|---|
| Nominal index (2010 = 100) | 93.2 | 99.8 | 109.0 | 120.2 | 132.2 | 2calculation |
| Nominal change | +3.1% | +7.1% | +9.2% | +10.3% | +10.0% | 2calculation |
| Real change | +2.5% | +6.5% | +7.8% | +9.2% | +9.6% | 1calculation |
| 12-month Euribor, annual average | 0.17% | −0.03% | −0.15% | −0.17% | −0.22% | 4calculation |
| Mortgage lending granted (€ million) | 5,516 | 7,726 | 9,568 | 10,268 | 10,331 | 17 |
| Home sales | — | 127,106 | 153,292 | 178,691 | about 181,400 | 8; 2017–2018secondary; 20199calculation |
| New homes permitted (family housing) | 8,618 | 11,811 | — | 21,896 | 24,905 | 9, 10 |
| Cumulative short-term rental registrations, year end | 24,343 | 35,606 | 54,503 | 77,053 (to 15 Oct.) | about 91,600 | 30; 2019secondary |
| Foreign population with a residence permit | 383,759 | 392,969 | 416,682 | 477,472 | 588,976 | 5 |
| Monthly minimum wage (€) | 505 | 530 | 557 | 580 | 600 | 22 |
For the new homes completed in this block, the official sources do not give the same number:
| Year | Number | Source |
|---|---|---|
| 2015 | about 6,600 | Implied by INE in 2017: 7,256 in 2016, 9.8% more than in 20158calculation |
| 2015 | 7,148 | INE, cited by Observador13 |
| 2015 | 7,817 | INE, Construction and Housing Statistics 20209 |
| 2017 | 8,213 | INE, Construction and Housing Statistics 202210 |
| 2019 | 12,954 | INE via PORDATAsecondary |
| 2019 | 13,092 | INE, Construction and Housing Statistics 20209 |
| 2015–2019, average | about 9,000 a year | +Liberdade14 |
The likely explanation lies in the method. For some years INE has estimated completed works with a regression model based on the deadline in the permit, and it revises earlier years11. Each edition of the report gives slightly different numbers for the same year. Whatever the version, 2015 and 2016 are the years with the fewest new homes completed this century: fewer than 8,000, against more than 100,000 a year in the early 2000s.
Permits recovered earlier than completions: new homes permitted tripled from 8,618 in 2015 to 24,905 in 20199. A permitted home takes years to be ready, so the new supply only arrived in the next block.
| Share of renovation | 2015 | 2016 | 2017 | 2018 | 2019 | Source |
|---|---|---|---|---|---|---|
| In buildings permitted | 30.7% | 29.8% | 26.4% | 25.1% | 24.0% | 9 |
| In buildings completed | 33.2% to 33.6% | 29.5% to 31.4% | 27.7% | 24.9% | 25.3% | 8, 9, 10 |
The highest share of renovation in completed works was in 2015, with a third of the total. Even the share itself varies between INE editions (33.2% or 33.6% for 2015). In Porto, reconstructions per 100 new constructions went from 8.8 in 2011–2013 to 31 in 2014–2016. In Lisbon they reached 12.8 in 2013–20158. The centres of both cities were rebuilt in this block.
Short-term rental registrations went from 14,127 at the end of 2014 to 77,053 in October 201830. At that date, the municipality of Lisbon had 15,881 registrations, Porto 6,972 and Albufeira 6,54230. In Lisbon, Airbnb listings went from about 2,000 in 2014 to 15,700 in 2018, 76% of them entire homes36.
The effect on prices was not the same everywhere. The Nova SBE study finds significant effects in Lisbon and Porto and no effect in the Algarve or the rest of the country38. In Lisbon and Porto, short-term rentals took homes that could have been permanent housing. The next chapter deals with this in detail.
The foreign population starts rising again from 2016 and accelerates in 2018: from 416,682 in 2017 to 588,976 in 2019, 41% more in two years5calculation. For 2017, AIMA gives 421,8026, 5,120 more than the PORDATA series. The difference is small, but it shows that even the starting point of this wave depends on the source.
Chapter 5 · Short-term rentals
In the historic centres of Lisbon and Porto, tourism did two things at once: it paid for the restoration of run-down buildings and it took those homes out of permanent housing. In Santa Maria Maior, between the 2011 and 2021 Censuses, 28% of family homes and 22% of residents disappeared31, 34. In the country as a whole, short-term rentals are too small to explain the price rise. The data supports the hypothesis in the neighbourhoods where the phenomenon was concentrated, and not outside them.
When tourism grows and short-term rentals become profitable, vacant or run-down buildings become worth restoring. Restoration is good for the neighbourhood: façades, roofs, structures. But the restored flats go into the tourist circuit and leave the housing market. The neighbourhood ends up restored and with fewer homes for those who want to live there.
The hypothesis has two parts that can be tested separately: (1) short-term rentals drove renovation; (2) the renovated homes left permanent housing.
The evidence in favour is consistent, from different sources.
The evidence against is mostly about scale. Renovation also rose as a national share, from 20.4% of completed works in 2008 to a third in 20157, 8, and that increase began before short-term rentals. That share rose mainly because new construction fell. Renovation driven by tourism is a historic-centre phenomenon, not a national one.
In Lisbon's central neighbourhoods, the data's answer is yes.
| Parish | Population 2011 | Population 2021 | Family dwellings 2011 → 2021 | Source |
|---|---|---|---|---|
| Santa Maria Maior | 12,822 | 9,997 (−22%) | 10,776 → 7,698 (−28%) | 34; change −27.8% in31 |
| Misericórdia | 13,044 | 9,645 (−26%) | 10,555 → 8,660 (−18%) | 34; −17.6% in31 |
| São Vicente | — | — | −6.2% | 31 |
| Santo António | — | — | −2.6% | 31 |
| Municipality of Lisbon | — | — | −1.9% | 31 |
There is a detail of method that strengthens the hypothesis: the Census does not count short-term rentals as family dwellings31. A home that became a short-term rental can simply drop out of the count. The 28% fall in Santa Maria Maior is consistent with this, although the Census cannot prove how many of those homes are short-term rentals today.
The Alfama study adds the mechanism36:
The short-term rental ratio in Santa Maria Maior is the clearest example of how the same fact gives very different numbers:
| Ratio | Value | How it is calculated | Source |
|---|---|---|---|
| Official (September 2022) | 52% | Short-term rentals over homes in the 2011 Census | 31 |
| 2011 base (City Council) | 51.7% | 2022 short-term rentals over homes in the 2011 Census | 33 |
| Without hostels, 2021 base | 60.7% | 4,669 short-term rentals over homes in the 2021 Census | 31 |
| City Council, 2021 base | 71.3% | 2022 short-term rentals over homes in the 2021 Census | 33 |
| With weighted hostels | 77.1% | Each hostel counts as several units | 31 |
| Residents' association (2026) | 69 per 100 occupied homes | National register crossed with the 2021 Census | 39 |
The change of base makes sense: as the 2021 Census counted fewer homes, the same number of short-term rentals weighs more. It is the phenomenon itself changing the ruler that measures it. In Misericórdia, the ratio goes from 39% to 51.8% depending on the formula31.
| Year | What happens | Source |
|---|---|---|
| 2008 | Short-term rentals appear in the law | diploma to be confirmed |
| 2009–2013 | First registrations in Santa Maria Maior: between 9 and 43 a year | 35 |
| 2014 | Dedicated short-term rental regime; 243 new registrations in Santa Maria Maior; 14,127 nationally | 30, 35, 38 |
| 2014–2018 | Short-term rentals grow about 100% a year in Lisbon | 34 |
| 2015 | Peak share of renovation in completed works (a third) | 8, 9 |
| 2015–2017 | In Alfama, 21 buildings renovated, 12 entirely for short-term rental | 36 |
| end of 2018 | First containment zones in Lisbon; works fall in those parishes | 32 |
| 2011–2021 | Santa Maria Maior loses 28% of its family homes and 22% of its residents | 31, 34 |
| 2025–2026 | Clean-up of inactive registrations; Lisbon goes from about 18,600 to 11,704 registrations | 39, 40 |
The hypothesis holds where short-term rentals were concentrated. In Lisbon's four central neighbourhoods, which account for 61% of the city's short-term rentals39, there was at the same time restoration of run-down buildings and an exit of homes and residents from housing. Nationally, short-term rentals are about 2% of the stock and are not enough to explain the general rise in prices. One question remains: without tourism, how many of these buildings would be restored today, and how many would still be derelict?
Chapter 6 · Money from abroad
Golden visas arrived in October 2012, in the middle of the troika, and brought about 6 billion euros into real estate up to 202262. In volume they weighed little: about 2% to 3% of the value of home sales in a typical yearcalculation. The strongest effect was concentrated in time and in segment: at the start of the rise (2013–2014), in homes around 500,000 euros and in the municipality of Lisbon. When the programme was restricted (2022) and closed to real estate (2023), prices kept rising.
| Year | Visas granted | Investment | Source |
|---|---|---|---|
| 2012 | 2 | — | 58 |
| 2013 | 494 | — | 58 |
| 2014 | 1,526 | about €921 million | 58; investmentsecondary |
| 2015 | 766 | — | 58 |
| 2016 | 1,414 | — | 58 |
| 2017 | 1,351 | — | 58 |
| 2018 | 1,409 | about €838 million | 58;59calculation |
| 2019 | 1,245 | €742 million | 59 |
| 2020 | 1,182 | €647 million | 60 |
| 2021 | 865 | about €461 million | 61;62calculation |
| 2022 | 1,281 | €654 million | 62 |
| 2023 | 2,091 | not published | 63 |
| 2024 | 2,081 | not published | 63 |
| 2025 | 2,685 (peak) | not published | 63 |
Up to 2022, 10,593 of the 11,535 visas were through property purchase, with 6,041 million of the 6,754 million euros invested (89%)62calculation. The renovation route added 1,485 visas and 534 million62. Job creation accounted for 22 visas in ten years62.
China dominated the early years: 4,467 of the 8,207 visas up to 2019, more than half59calculation. In 2025 the United States led for the fourth year running (535), ahead of China (392) and the United Kingdom (252)63.
| Up to | Visas | Investment | Source |
|---|---|---|---|
| December 2022 | 11,535 | €6,754 million (total); €6,041 million (property) | 62 |
| January 2023 | 11,628 | about €6.7 billion (total); about €6 billion (property) | 26 |
| September 2023 | 12,718 | €6,451 million | 63 |
| 2025 | 18,254 | not published | 63 |
The September 2023 figure is lower than the December 2022 one. The most likely explanation is that the 6,451 million cover only the property route; other sources give 7,318 million in total up to that dateto be confirmed. Since the move from SEF to AIMA, there is no longer a public investment figure for 5,536 visas63.
In volume, it was small. In 2019, the 661 million euros of visas through property purchase59 were equivalent to about 2.6% of the value of home sales that year (about 25.6 billion)9calculation. In number, 1,160 property visas against about 181,000 sales: 0.6%9, 59calculation. In 2022, total visa investment was about 2.1% of the value of sales10, 62calculation.
In the expensive segment, it was strong. In 2019, the average investment per visa on the 500,000-euro route was about 618,000 euros (584.5 million for 946 visas), more than four times the average value of a home sale (about 141,000 euros)9, 59calculation. The most complete study, by Pereira dos Santos and Strohmaier (IZA, 2024), uses every deed from 2007 to 2019 and finds67:
In timing, it coincided with the turn. The 2014 visa peak (1,526) coincides with the highest share of non-residents in the value of sales (15.7%) and with the widest price gap between non-residents and residents (2.38 times)58, 69. Money from abroad came in strongly exactly when Portuguese credit was at its lowest and prices at the bottom.
But it was not the engine of the national rise. Prices kept rising in 2015, when visas halved; in 2022, after the restriction; and they accelerated in 2025 (+17.6%), two years after the end of the property route2, 58, 63. The association Transparência e Integridade blames the visas for "the dizzying rise in house prices"68, but presents no numbers to support it.
Yes, golden visas boosted the sale of expensive property: they concentrated demand at 500,000 euros, pushed up prices in that segment by more than 10%, and brought in money from abroad at a moment when nobody else was buying. Nationally, they were a small fraction of sales, and the price rise survived their restriction and their end. The role the data supports is that of a fuse in central Lisbon between 2013 and 2016, not the fuel of the decade.
Crossing prices with sales to non-residents gives two results that seem contradictory and are not. Over time, the less buyers from abroad weigh, the more prices rise. On the map, the regions where buyers from abroad buy most are among the most expensive. Foreigners helped set the level of prices in certain places and at the start of the cycle; the recent acceleration comes from buyers who live in Portugal.
Between 2014 and 2025, the share of non-residents in the value of sales fell from 15.7% to 8.2%, and the annual rise in prices went from 4.3% to 17.6%2, 9, 69calculation. Over the eight years with data, the correlation between the two series is −0.85; without 2025, −0.68calculation. Three caveats: there are only eight points; the scope changes halfway (to 2020 INE counts all property, afterwards only housing); and a correlation does not say what causes what.
What the series suggests: the peak share of buyers from abroad was in 2014, at the bottom of the market, when Portuguese credit was at its lowest and golden visas at their peak58, 69. As resident buyers came back (with credit, low interest rates and, later, immigration and support for young buyers), the share of buyers from abroad was diluted and prices accelerated.
| Year | Non-residents pay, on average, how many times the residents' price | Measure | Source |
|---|---|---|---|
| 2012 | 1.80 | average price per property, all property | 69 |
| 2014 | 2.38 (peak) | same | 69 |
| 2017 | 1.56 | same | 69 |
| 2019 | 1.66 | same | 69 |
| 2023 | about 1.79 | average price per home | 70calculation |
| Q2 2026 | 1.31 | median price per m² | 47calculation |
The gap was widest in 2014 and narrowed afterwards. Between 2012 and 2019, the average price paid by residents grew 7.7% a year, more than that paid by non-residents (6.8%)69. From a certain point, it was buyers from inside who pushed prices up more. The measures in the table are not the same (average per property, average per home, median per square metre): they show the trend, not numbers to compare one by one.
| Region | Share of non-residents in the value of sales | Median price (€/m²) | Sources |
|---|---|---|---|
| Algarve | 38.7% (2019); 38.5% (2023) | 3,139 (2025) | 69, 70, 54 |
| Oeste | 18.0% (2019) | 2,220 (Q1 2026) | 69, 72 |
| Alentejo Litoral | 15.0% (2019) | 2,150 (Q1 2026) | 69, 72 |
| Lisbon Metropolitan Area | 10.4% (2019) | 3,439 (Grande Lisboa, 2025) | 69, 54 |
| Norte | 3.1% (2019–2021) | 2,305 (Porto Metropolitan Area, 2025) | 71, 54 |
| Portugal | 13.3% (2019); 8.2% (2025) | 2,076 (2025) | 69, 542025 secondary |
The regions where buyers from abroad weigh most (Algarve, Oeste, Alentejo Litoral) are above the national median, and the Algarve is the second most expensive region in the country. The exception is Grande Lisboa: it is the most expensive, with a much smaller share of non-residents than the Algarve. In Lisbon, the price is set mainly by those who live there or want to. In the Algarve, close to 40% of the value comes from those buying from abroad. The years in the two columns do not match; the INE table by sub-region with share and price for the same year, needed to make this comparison rigorously, is missing.
Chapter 7 · 2020–2025
In 2025 the price index rose 17.6%, the biggest annual rise of the period, and the value of sales reached 41.2 billion euros, the highest in the series that began in 20092, 55. In this block everything moved: population, interest rates, credit, construction, rents and rules. It is the block in which it is hardest to separate the causes.
| Variable | 2020 | 2021 | 2022 | 2023 | 2024 | 2025 | Source |
|---|---|---|---|---|---|---|---|
| Nominal index (2010 = 100) | 143.7 | 157.3 | 177.1 | 191.7 | 209.1 | 245.8 | 2calculation |
| Nominal change | +8.8% | +9.4% | +12.6% | +8.2% | +9.1% | +17.6% | 2calculation |
| Real change | +8.8% | +8.0% | +4.5% | +3.7% | +6.5% | +14.9% | 1calculation |
| Implied inflation | 0.0% | 1.3% | 7.8% | 4.3% | 2.4% | 2.3% | 1, 2calculation |
| 12-month Euribor, annual average | −0.30% | −0.49% | 1.09% | 3.87% | 3.27% | 2.22% | 4calculation |
| New homes completed (family housing) | 16,710 | 19,565 | 20,156 | 23,100 to 23,652 | 25,300 | 26,700 to 27,301 | 9, 10, 15, 16; 2025 highsecondary |
| New homes permitted | 25,083 | 29,312 | 30,247 | 32,519 | 34,600 | 41,900 | 9, 10, 15, 16 |
| Home sales | 171,800 | — | 167,900 | — | about 156,400 | 169,812 | 9, 10, 55; 2024calculation |
| Median rent of new contracts (€/m²) | 5.61 | — | 6.52 | — | 7.97 | 9.29 (new method) | 9, 10, 48; 2025to be confirmed |
| Foreign population (PORDATA series to 2023, AIMA in 2024) | 661,607 | 698,536 | 781,247 | 1,044,238 | 1,543,697 | — | 5, 6 |
| Monthly minimum wage (€) | 635 | 665 | 705 | 760 | 820 | 870 | 22 |
The 12-month Euribor went from −0.49% in 2021 to 3.87% in 20234calculation. In 2022 and 2023 real price growth slowed to 4.5% and 3.7%1calculation, and sales fell. But nominal prices never fell: they rose 12.6% and 8.2% in those two years. When rates came down, in 2024 and 2025, prices accelerated to +9.1% and +17.6%.
The interest-rate brake slowed the price rise without stopping it. It is an important difference from 2008, when rising rates coincided with falling prices.
Banco de Portugal publishes mortgage lending in series that are not comparable with each other (with and without renegotiations, for example). For 2023 there are three numbers: 18.2 billion euros, including renegotiations18; 24.5 billion, in a thesis citing Banco de Portugal19; and about 13.1 billion in new contracts, according to other estimatesto be confirmed.
With that caveat, the direction is clear. In 2021, 14.8 billion were lent17. In 2025, for permanent own homes alone, it was more than 19 billion in more than 105,000 contracts45. The stock of mortgage lending went from 87.4 billion in 2019 to 101.3 billion in 202117.
Two groups gained weight in new lending:
The foreign population went from 661,607 in 2020 to 1,044,238 in 2023 and 1,543,697 in 20245, 6. In 2024, Brazilians were 31.4%, Indians 7.4%, Angolans 6.9% and Ukrainians 5.9%6.
Three caveats before linking this number to prices:
Median rents for new contracts went from €5.61/m² in 2020 to €7.97/m² in 2024, +42% in four years9, 48calculation. Overcrowded dwellings rose 17.1% between 2011 and 202112. A Banco de Portugal study (October 2026) estimates that a 1% increase in housing demand raises prices 0.69% and rents 0.37%, more in municipalities where supply responds less to price42. In that study's model, Lisbon, Sintra, Gaia and Porto appear among the municipalities where supply responds more, not less (see chapter 9, point 10).
| Measure | Value | Source |
|---|---|---|
| Buyers with tax residence outside Portugal, 2020 | 8.3% of homes, 14.0% of value | INE9 |
| Buyers with tax residence outside Portugal, 2025 | 5.0% of homes, 8.2% of value | INEsecondary |
| Buyers of foreign nationality, resident or not, 2019 | 25% | Banco de Portugal43 |
| The same, 2023 (peak) | 31% | Banco de Portugal43 |
| The same, 2025 | 28% | Banco de Portugal43 |
The two numbers measure different things. INE counts those who do not live in Portugal. Banco de Portugal counts those without Portuguese nationality, whether they live here or not. The difference (from 5% to 28%) is mostly foreigners who live in Portugal and buy a home, a group that grew with immigration (an inference from the two definitions; the series separating foreign residents is missing). Non-residents lost weight; foreign residents gained it.
They pay more: according to Banco de Portugal, foreign residents and non-residents pay on average 43% more per home44. In Q2 2026, the median price paid by non-residents was €3,182/m², against €2,421/m² paid by residents47.
New homes completed went from 16,710 in 2020 to about 27,000 in 2025, and permits to 41,900, the highest level since 20089, 16. Even so, 2025 construction is about a quarter of the 2000–2004 average14calculation. The sector says capacity "is at the limit", with a shortage of 80,000 to 100,000 workers49.
Banco de Portugal estimates an accumulated deficit of about 300,000 homes over the last decade: demand (mainly population and household growth through immigration) grew more than homes completed. The governor considers that, in 2026, the annual flow of supply and demand has already balanced, but that the accumulated deficit remains16.
The gap between the metrics is 40 percentage points or more in the Portuguese case29. The price-to-income and price-to-rent ratios give very high values; the econometric model, which also considers interest rates and demand, gives lower values (KBC's reading).
Chapter 8 · Geography
In 2025, a square metre sold in Lisbon cost 4,875 euros, more than double the national median of 2,07654. But in 2026 the fastest rise is no longer in Lisbon: it is in Médio Tejo (+25.4%), Oeste (+24.0%) and Lezíria do Tejo (+23.9%), while Lisbon and Porto slow down47. The pressure that began in the historic centres spread outwards in rings.
| Region | €/m² | Against the national median |
|---|---|---|
| Grande Lisboa | 3,439 | +66% |
| Algarve | 3,139 | +51% |
| Península de Setúbal | 2,596 | +25% |
| Madeira | 2,500 | +20% |
| Porto Metropolitan Area | 2,305 | +11% |
| Portugal | 2,076 | — |
Source: INE, Construction and Housing Statistics 2025, via Observador54; last columncalculation. The most expensive municipalities are Lisbon (4,875), Cascais (4,550), Oeiras (4,187), Loulé (3,993) and Lagos (3,801)54.
In Q2 2026, prices rose in all 26 sub-regions of the country, with the national median at €2,445/m² (+18.4%)47. Lisbon, at €5,299/m², slowed by 8.9 percentage points; Porto slowed by 5.4 points. The biggest rises were in the regions around the Lisbon Metropolitan Area: Médio Tejo, Oeste, Lezíria do Tejo and Península de Setúbal (+22.2%)47.
The pattern is consistent with people leaving Lisbon because they cannot buy or rent there. The data gathered here does not allow that movement of people to be proven: it is a hypothesis to test with INE's internal migration data.
| Region (Q2 2026) | Non-residents (€/m²) | Residents (€/m²) | Difference |
|---|---|---|---|
| Grande Lisboa | 5,878 | 4,000 | +47% |
| Porto Metropolitan Area | 3,640 | 2,648 | +37% |
| Madeira | 3,611 | 2,785 | +30% |
| Algarve | 4,143 | 3,368 | +23% |
| Península de Setúbal | 3,240 | 3,063 | +6% |
Source: INE via DN47; Lisbon and Porto differences as published, the restcalculation. Non-residents pay more per square metre, but they may also buy different homes (more central, newer or renovated). The median does not separate the two, and I did not find a comparison for identical homes.
In 2024, the median rent of new contracts was €15.93/m² in the municipality of Lisbon and €12.58/m² in Porto, against €7.97/m² nationally48. In Lisbon, rent per square metre is double the national figure.
The study by João Amador (Banco de Portugal, October 2026) concludes that demand shocks have "substantially larger" effects where supply responds less to price. Contrary to how the press summarised it, the study does not put Lisbon and Porto at the top of those municipalities: by its measure, Lisbon has the highest supply elasticity in the country and Sintra the smallest effect of demand on prices. In the Algarve, yes: supply that responds little meets demand from non-residents and investment42. A study by the consultancy Neoturis says Lisbon had only 2,352 new homes between 2015 and 202340secondary.
| Region (2021 Census) | Vacancy rate |
|---|---|
| Alto Alentejo and Médio Tejo (highest) | 17.4% |
| Portugal | 12.1% |
| Algarve | 11.9% |
| Grande Lisboa | 11.1% |
| Porto Metropolitan Area | 10.0% |
Source: INE12. Empty homes exist everywhere, but they weigh more where demand is lower. Even so, Grande Lisboa has 11.1% vacant homes: Lisbon's problem is not only the number of homes, it is also homes that are not on the market.
The country has at least four housing markets, and each moves for different reasons. The portraits follow the same outline: what moves the price, who buys, how much is built, which explanations compete and where the numbers do not add up. The common basis is Banco de Portugal's June 2026 Economic Bulletin, which crosses prices, rents, short-term rentals and credit municipality by municipality73.
| Region | Median price 2025 (€/m²) | Non-resident premium (Q2 2026) | Weight of short-term rentals | What seems to weigh most |
|---|---|---|---|---|
| Lisbon (municipality) | 4,875 | +47% (Grande Lisboa) | High in the historic centre; 6,765 registrations ordered cancelled in 2026 | Outside demand and lack of land |
| Porto (metropolitan area) | 2,305 | +37% | About 12% of the Porto municipality's registrations to be cancelled | Tourism and renovation in the centre, then the metropolitan ring |
| Algarve | 3,139 | +23% | More than 10% of homes in eight municipalities | Second homes and non-resident investment |
| Lisbon's periphery (Península de Setúbal) | 2,596 | +6% | Low, below 1.5% in most peripheral municipalities | Resident families leaving the centre |
Sources: prices54; non-resident premium47, with Lisbon and Porto as published and the restcalculation; short-term rentals73, 74.
What moves the price. Lisbon has the most expensive square metre in the country (4,875 euros in 2025) and a rent for new contracts that is double the national figure54, 48. Banco de Portugal shows why supply responds poorly: in the metropolitan areas urban land use is more intense, less land is left, and old, low-density buildings make renovation and densification harder73. The housing that exists is also poorly distributed: in 2021, 24% of homes with four or more rooms in Lisbon had a single occupant, against 18.8% nationally73.
Who buys. Lisbon was the municipality with the most homes sold in 2025 (8,235)75. In Grande Lisboa, buyers with tax residence abroad paid 47% more per square metre than residents in Q2 202647.
How much is built. Little. The consultancy Neoturis counts 2,352 new homes in the city between 2015 and 202340secondary. Among completed homes, small types (studio to two-bedroom) gain weight in Lisbon, Porto and the Algarve73.
Short-term rentals. In February 2026, the City Council ordered 6,765 registrations cancelled, about 40% of the 19,000 in existence, for lack of liability insurance74. The association Vizinhos em Lisboa notes that the homes did not return to the housing market39.
Competing explanations. The supply reading says Lisbon built far too little for the households its area gained (see the regional test in this chapter). João Amador's study adds a nuance: between 2017 and 2024, Lisbon's supply was, in percentage terms, among those that reacted most to price42. Both can be true at once: reacting strongly from almost nothing is still little. The outside-demand reading points to the 47% non-resident premium. The two are not mutually exclusive: Banco de Portugal concludes that prices and rents rose together, which fits a market driven by fundamentals rather than by a bubble of expectations73.
Where it does not add up. The short-term rental clean-up cut about 40% of registrations, but no effect is visible on the supply of homes. Either the cancelled registrations were already inactive ("ghosts"), or the homes went to another use. Public data does not allow the two hypotheses to be separated. The 2026 slowdown (8.9 percentage points less47) also has two readings: Lisbon has reached the limit of what people will pay, or demand has moved out.
What moves the price. The Porto Metropolitan Area has a median price of €2,305/m², only 11% above the national figure54. The municipality of Porto is something else: the rent of new contracts was €12.58/m² in 2024, more than 50% above the national figure48calculation. The sequence repeats Lisbon with a delay: historic centre renovated, tourism, short-term rentals, then the rise spreading through the ring.
Who buys. Non-residents paid 37% more per square metre than residents in Q2 202647. In 2025, Vila Nova de Gaia sold more homes (5,494) than Porto itself (4,503)75: the metropolitan area already trades more outside the city than inside it.
Short-term rentals. In May 2026, Porto City Council ordered 1,413 registrations cancelled, about 12% of the total74, a much smaller clean-up than Lisbon's.
A trait of its own. The Porto Metropolitan Area is among the areas with the most social housing in the country73. Banco de Portugal credits social housing with a stabilising effect on rents73. Whether that explains part of the difference from Lisbon is a hypothesis, not a result: the data gathered here does not test it.
Where it does not add up. In Q1 2026, Porto's rate of increase accelerated by 2.4 percentage points54. In Q2, it slowed by 5.4 points47. Two consecutive quarters with opposite signs show that the quarterly municipal series is volatile: you cannot draw a trend from one quarter.
What moves the price. The Algarve has the second-highest regional price (€3,139/m², 51% above the national median), with Loulé (3,993) and Lagos (3,801) among the five most expensive municipalities in the country54. It is also where short-term rentals weigh most: in eight municipalities they take more than 10% of homes, against 1.6% on the mainland73.
Who buys. Banco de Portugal writes that price-to-income ratios in the Algarve are above the national average and strongly influenced by non-resident demand73. According to INE, non-residents made 20.1% of home purchases in the Algarve in 2025 and 31.3% of the value, against 28.5% and 42.4% in 201981calculation.
How much is built. Banco de Portugal identifies in the Algarve the same urban pattern as in Lisbon and Porto: medium-height buildings and small homes (studio to two-bedroom), linked to density, land costs and tourism73.
Competing explanations. The tourism reading says the Algarve is paid for at prices from abroad. The supply reading says that even there construction did not keep up with demand. João Amador's study shows that the Algarve weighs on the national results: when the region is isolated, the effect of 1% of demand on prices falls from 0.69% to 0.51%, because there supply that responds little meets demand from non-residents and investment42. Banco de Portugal adds a note of scale: international studies find significant effects of short-term rentals on prices, but they explain only a minority of the rise73.
Where it does not add up. The non-resident premium is the lowest of the large regions (+23%47), even though the weight of foreigners is the highest. One possible explanation: when the buyer from abroad is a third of the market's value, they stop paying "more" than the others and start setting the price for everyone. Another sign pointing the same way: Algarve residents expect the smallest price rise next year (3.7%, against 6.6% in the Lisbon Metropolitan Area)73. Those who live there are not the ones pushing the price.
What moves the price. Between 2017 and 2025, prices more than doubled in 157 of the 286 municipalities with data. In Sintra, Seixal, Barreiro, Moita and Setúbal, the rise per square metre passed 200%73. In rents, Moita is among the three municipalities where the rent of new contracts rose more than 125% between 2017 and 202473. Banco de Portugal calls this the "territorial spread of appreciation": demand moved to municipalities where, in 2017, buying was cheap compared with renting73.
Who buys. Mainly residents. In Península de Setúbal, non-residents paid only 6% more than residents47. Sintra was the second municipality in the country by number of homes sold in 2025 (6,363)75. Short-term rentals are residual73.
The role of credit. You would expect the periphery, where families buy with a loan, to have risen on the back of credit. Banco de Portugal does not find that relationship: between 2019 and 2025, the municipalities with the biggest price rises did not have a larger increase in the weight of credit or in the loan-to-value ratio73.
The wave continues. In Q2 2026, the biggest rises in the country were around the Lisbon Metropolitan Area: Médio Tejo, Oeste, Lezíria do Tejo and Península de Setúbal (+22.2%)47.
Where it does not add up. In percentage terms, the periphery led. In euros, the distance to Lisbon remains: €2,596/m² in Península de Setúbal against 4,875 in the municipality of Lisbon54. A rise of 200% from a low base does not make the periphery an expensive market; it makes it a market that stopped being cheap. And the most obvious explanation, families leaving Lisbon, is still to be proven with internal migration data.
The direction is the same everywhere; the engine is not. In Lisbon and the Algarve, outside demand and the lack of land where demand is concentrated weigh most. In Porto, the Lisbon pattern arrived later and looks more moderate. In the periphery, buyers are mainly residents, and the rise largely measures Lisbon's price overflowing. To the question "who sets the price?", the answer changes with the map: a single thesis for the whole country risks describing one region well and the other three badly.
The thesis says the housing shortage is local. If that is true, it has to show in the numbers: in the most sought-after regions, the number of new homes has to fall far short of the number of new households, and in the rest of the country it does not. The data came directly from INE's database: new homes completed by region since 199576, households in the 2001, 2011 and 2021 Censuses77, 78 and homes by type of occupancy in the three Censuses79.
| Region | New households (Census 2011–2021) | New homes completed (2011–2020) | New homes per new household |
|---|---|---|---|
| Grande Lisboa | +24,322 | 13,363 | 0.55 |
| Península de Setúbal | +20,883 | 9,561 | 0.46 |
| Grande Porto | +22,830 | 12,737 | 0.56 |
| Algarve | +11,032 | 8,852 | 0.80 |
| The four regions | +79,067 | 44,513 | 0.56 |
| Rest of the country | +27,159 | 86,515 | 3.19 |
| Portugal | +106,226 | 131,028 | 1.23 |
Sources:76, 77, 78; differences and ratioscalculation. The regions are the 2002 NUTS III (Grande Lisboa: the nine municipalities north of the Tagus; Grande Porto: the nine central municipalities of the metropolitan area). In 2011 INE counted "classic families" and in 2021 "private households", both in homes of usual residence: the concepts are close but not identical. The Lisbon Metropolitan Area's own assessment, using another source, reaches almost identical values (+24,046 in Grande Lisboa, +21,370 in Península de Setúbal)80.
The four regions gained three quarters of the country's new households and received a third of its new homes. Outside them, more than three homes were built for each new household. The country as a whole built more homes than it gained households; the problem is where.
| Municipality | New households (2011–2021) | New homes (2011–2020) | Homes per new household |
|---|---|---|---|
| Lisbon | −927 | 1,735 | — |
| Sintra | +9,046 | 1,732 | 0.19 |
| Vila Nova de Gaia | +7,028 | 2,788 | 0.40 |
| Seixal | +5,335 | 2,908 | 0.55 |
| Porto | +2,099 | 2,903 | 1.38 |
Sources:76, 77, 78;calculation. Lisbon completed 1,735 new homes in a decade, fewer than 200 a year, and lost households. Sintra, the second municipality in the country for home sales75, gained more than nine thousand households with fewer than two thousand new homes. Porto is the exception: it is the only one of these municipalities that built more homes than it gained households.
| Region | New homes 2001–2010 | 2011–2020 | 2021–2025 | Worst year against the peak |
|---|---|---|---|---|
| Grande Lisboa | 95,578 | 13,363 | 14,690 | 473 in 2016 (2.6% of the 2000 peak) |
| Península de Setúbal | 59,285 | 9,561 | 9,604 | 352 in 2014 (3.4%) |
| Grande Porto | 98,006 | 12,737 | 20,280 | 455 in 2015 (2.3%) |
| Algarve | 78,706 | 8,852 | 6,731 | 536 in 2015 (5.2%) |
| Portugal | 762,594 | 131,028 | 115,220 | 7,148 in 2015 (5.7%) |
Source:76; the 2021 to 2025 values are provisional at INE. In their worst year, Grande Lisboa and Grande Porto were building between 2% and 3% of what they built at the peak; the country stayed at 5.7%. After 2021 the regions diverge: Grande Porto went from 9.7% to 17.6% of the country's new homes, while the Algarve fell from 10.3% (2001–2010) to 6.8% and then 5.8%, despite being the region that has gained the most population since 2021 (+13.8%, in INE's new series)76secondary for population.
| Region | Vacant homes 2001 | 2011 | 2021 | Second homes 2001 | 2011 | 2021 |
|---|---|---|---|---|---|---|
| Grande Lisboa | 11.9% | 12.8% | 11.1% | 10.7% | 10.1% | 8.5% |
| Península de Setúbal | 10.9% | 11.7% | 9.6% | 17.5% | 15.2% | 12.8% |
| Grande Porto | 11.9% | 12.7% | 10.1% | 9.7% | 9.5% | 8.4% |
| Algarve | 9.4% | 13.2% | 11.9% | 38.5% | 39.5% | 38.6% |
| Portugal | 10.8% | 12.5% | 12.1% | 18.4% | 19.3% | 18.5% |
Source:79; percentages of total homescalculation. Between 2011 and 2021, vacant homes fell by 44,653 in the four regions and by only 11,913 nationally: outside them, they rose by about 33,000calculation. In Grande Lisboa, Península de Setúbal and Grande Porto, second homes also fell, by about 32,000 in total. The demand that found no new home went into homes that already existed: vacant homes occupied again and second homes that became family homes. Not in the Algarve: the share of second homes stayed at 39%.
This also changes how to read the 723,000 empty homes of 2021. In 2001, the country already had 543,777 vacant homes (10.8%)79. Vacancies grew in the decade when a lot was built and fell, in the sought-after regions, in the decade when building stopped.
INE's quarterly data by the buyer's tax residence81 confirms the weight of buyers from abroad in the Algarve and shows it is falling:
| Algarve | Purchases by non-residents (number) | Share of value | Average value, non-resident | Average value, resident |
|---|---|---|---|---|
| 2019 | 28.5% | 42.4% | €298,000 | €162,000 |
| 2022 | 26.2% | 37.9% | €411,000 | €238,000 |
| 2025 | 20.1% | 31.3% | €572,000 | €316,000 |
Source:81; sum of the four quarters and difference between the total and purchases with tax residence in Portugalcalculation. The average value per purchase is not price per square metre: it mixes sizes and types of home.
| Region | Median price 2025 (€/m²) | Average monthly earnings 2024 | Months of salary per m² | Years of salary for 90 m² |
|---|---|---|---|---|
| Algarve | 3,139 | €1,320 | 2.4 | 15.3 |
| Grande Lisboa | 3,439 | €1,936 | 1.8 | 11.4 |
| Península de Setúbal | 2,596 | €1,545 | 1.7 | 10.8 |
| Porto Metropolitan Area | 2,305 | €1,620 | 1.4 | 9.1 |
| Portugal | 2,076 | €1,576 | 1.3 | 8.5 |
Sources: prices54; average monthly earnings of employees, Quadros de Pessoal82;calculation, with 14 payments a year and before tax. The Algarve is the region where a home is least affordable for those who work there: the price is set by those who buy with income from abroad. Nationally, average earnings rose 44% between 2015 and 2024 (from 1,094 to 1,576 euros)82, while the house price index rose more than double that2calculation.
Chapter 9 · Contradictions
On a subject where everyone quotes numbers, the official sources disagree on almost every key quantity: how many homes are built, how many immigrants there are, how much credit is granted, how many homes are missing, how much the foreign buyer weighs. The differences are not mistakes: they come from definitions, revisions and choices of method. But they change the conclusion you draw. These are the eleven most important contradictions, each with what explains it and what it changes.
| Year | Numbers found | Sources |
|---|---|---|
| 2013 | 20,689 (2014 edition) or 12,751 (current series in PORDATA) | 7; PORDATAsecondary |
| 2015 | about 6,600, 7,148 or 7,817 | 8calculation;13;9 |
| 2019 | 12,954 or 13,092 | PORDATAsecondary;9 |
| 2023 | 23,100 or 23,652 | 16;15 |
| 2025 | 26,700 or 27,301 | 16; INEsecondary |
What explains it: INE does not count completed homes one by one; it estimates them with a regression model based on the deadline in the permit, and revises earlier years11. The biggest difference, for 2013 (20,689 against 12,751), suggests a deep revision of the series, still to be documented.
What it changes: the order of magnitude holds (construction fell to a tenth and recovered to a quarter of the early 2000s). But any calculation of "missing homes" made with one version can differ by tens of thousands from one made with another.
| Year | Numbers found | Sources |
|---|---|---|
| 2017 | 416,682 or 421,802 | 5;6 |
| 2022 | 781,247 or 994,011 (revised) | 5;to be confirmed |
| 2023 | 1,044,238 or about 1.29 to 1.30 million (revised) | 5;to be confirmed |
| 2024 | 1,543,697 | 6 |
What explains it: AIMA began counting pending regularisation cases, and earlier series were revised upwards. INE also revised net migration and the resident populationJune 2026 revision to be confirmed.
What it changes: "quadrupled in seven years" (from 421,802 to 1,543,697) compares an old count with a new one. With the revised series, the rise is still huge, but the year-on-year pace changes.
| Year | Numbers found | Sources |
|---|---|---|
| 2012 | about 1.9 billion or 2.03 billion | 19;17calculation |
| 2019 | 10.33 billion or about 10.95 billion | 17;19calculation |
| 2023 | 13.1, 18.2 or 24.5 billion | to be confirmed;18;19 |
What explains it: Banco de Portugal has series with and without renegotiations, and a series of new contracts that only begins in December 2014. In 2023, with rates rising, many families renegotiated, and that inflates the series that include them.
What it changes: saying that 2023 lending beat 2007's (19.6 billion) depends on the series chosen.
Almost 240%27, 180%secondary, 175% or 164%2calculation, or 114% in real terms1calculation. See "Before you read: how a home's price is measured".
About 35% (European Commission, data to 2024)28, +40% (average of four ECB metrics, early 2026)29, or below 20% (ECB econometric model, according to KBC)29. In 2007, Banco de Portugal saw no overvaluation at all20.
What it changes: at 35% to 40%, the risk of a correction is high. At under 20%, prices are expensive but explained by fundamentals.
| Estimate | Who | Method | Source |
|---|---|---|---|
| 150,000 to 200,000 | Developers (APPII and others) | Not explained | 49 |
| 180,000 over the next decade | Study cited by idealista | To be confirmed | secondary |
| About 300,000 (2021–2025) | Financial Times, Banco de España method | New households against homes permitted two years earlier | 50 |
| About 300,000 (last decade) | Banco de Portugal | New supply against what was needed to keep up with demand | 16 |
| 380,000 | Construction sector | To be confirmed | secondary |
| 723,215 vacant homes | 2021 Census | Count on Census day | 12 |
What explains it: "missing" depends on where, for whom and at what price. Vacant homes are mostly where demand is not: the vacancy rate is 17.4% in Alto Alentejo and Médio Tejo and 11.1% in Grande Lisboa12. But even Lisbon has 48,018 vacant homes39.
What it changes: the two statements "there are not enough homes" and "there are homes to spare" are true at the same time, in different places.
5.0% (INE, tax residence outside Portugal, 2025) or 28% (Banco de Portugal, foreign nationality, 2025)43secondary. See chapters 6 and 7.
About 125,000 registrations in 2025, more than 40,000 of them inactive40. In Lisbon, about 18,600 registrations before the clean-up and 11,704 after39, 40. In Santa Maria Maior, a ratio between 52% and 77% of homes, depending on the formula31, 33.
What it changes: old registrations inflated the size of short-term rentals for years, and the containment rules were decided with those numbers40.
The median rent of new contracts was €7.97/m² in 202448. For 2025, INE publishes €9.29/m² (+9.7%) with a new methodologyto be confirmed. If the 9.7% change is applied to 2024's 7.97, it gives about 8.74, not 9.29calculation: the new method changed the level of the series. For Lisbon, there is €15.93/m² for the municipality in 202448 and 13.06 for Grande Lisboato be confirmed.
The press summarised João Amador's study by saying that the municipalities with the least supply, with Lisbon and Porto at the top, have the biggest rises. The original study says something else42. The supply elasticity it estimates (how much the housing stock grows when the price rises, between 2017 and 2024) goes from 0.5% in Lisbon, the highest in the country, to negative values in Lajes das Flores, Alvito and Mourão. The effect of 1% of demand on prices goes from 0.4% in Sintra to 0.8% in Mourão, with Lisbon, Faro, Porto and Gaia among the lowest. The author himself warns that these elasticities are simple correlations, not causal effects, and that they are correlated with population density (0.49).
The two measures answer different questions. The study's asks how much supply reacts to price; my regional test asks how many new homes there were for each new household. Lisbon can react strongly in percentage terms and still build far fewer homes than the households it gains. What cannot be written is that João Amador's study shows Lisbon and Porto as the municipalities with the most rigid supply.
| Source | Measure | Value |
|---|---|---|
| IHRU, Report on Residential Renting, 2023 | Surplus of homes over households (stock) | 1,821,581 homes too many92secondary |
| National Housing Programme | Homes per household in 2021 | 1.4491 |
| INE/LNEC, 2024 | Quantitative shortfall in 2021 | 136,800 homes12 |
| Banco de Portugal, June 2026 | Construction deficit against new households, last decade | about 300,000 homes73 |
| This analysis | New homes per new household, 2011–2020, in the four regions | 0.56 (3.19 in the rest of the country) (chapter 8) |
IHRU concludes that "in no municipality are there more households than dwellings" and that the problem is not in the existing stock93secondary. Banco de Portugal and this analysis measure flows and reach the opposite. They neither add up nor reconcile: they are different questions. IHRU itself gives a clue for bringing them together: of the habitable vacant homes, 248,534 are off the market, mainly because the occupant died or the estate is awaiting division, with Lisbon (21,801) and Sintra (8,196) at the top93secondary. A home that exists but is neither for sale nor for rent is not supply. João Amador's study follows the same logic: it excludes from supply the vacant homes in regions losing population, "since these cannot meet demand pressures in growing regions"42.
Chapter 10 · Synthesis
No factor explains the 25 years on its own. Credit explains the 2008–2013 fall well and the 2001–2007 rise badly. Short-term rentals explain the historic centres well and the country badly. Immigration has weighed on rents since 2018 and did not weigh on prices in 2005–2009. The only factor present in every block is the relationship between new supply and demand shocks.
| Factor | 2000–04 | 2005–09 | 2010–14 | 2015–19 | 2020–25 |
|---|---|---|---|---|---|
| New construction (supply) | holds prices back | holds prices back | irrelevant (no demand) | amplifies | amplifies, less |
| Credit and interest rates | strong, no effect on prices | strong until 2007, then a brake | decisive in the fall | moderate | strong, both ways |
| Wages and income | weak | weak | against (falling) | weak | moderate |
| Tourism and short-term rentals | — | — | start (2014) | strong in the historic centres | weak nationally, strong locally |
| Investor and non-resident buyer | weak | weak | start (2012) | moderate | falling |
| Immigration | weak | strong in number, no effect on prices | against (people leave) | moderate | strong, mainly on rents |
| Public policy | end of subsidised credit | — | seeds (visas, rents, renovation) | weak brakes | demand support (young buyers) |
This table is a reading summary, not a statistical model. Each cell rests on the data in the blocks and is there to compare periods, not as an exact measure.
For: Banco de Portugal concludes that the housing problem does not come from a general lack of urban land, but from the concentration of demand in areas where land is scarce73. Vacant homes point the same way: the rate is highest where demand is lowest (17.4% in Alto Alentejo and Médio Tejo)12. Between 2017 and 2025, prices more than doubled in 157 municipalities, with the biggest rises in Lisbon's periphery73.
Against: Grande Lisboa has 11.1% vacant homes, almost the national average12: even where demand is tight there are homes off the market. And in 2026 the fastest rises are in sub-regions such as Médio Tejo and Lezíria, where land is not lacking47.
Reading: the housing shortage is local, not national. This resolves a good part of the contradiction between the estimated deficit and the 723,000 empty homes: many of the empty ones are where nobody is looking. Demand that does not fit in the centre spreads outwards in waves, and the price follows.
Across the five blocks a sequence repeats: when construction was high, demand and credit shocks did not pass into prices; since construction fell to a tenth and then a quarter of its peak, all of them have. The question this reading leaves open is why supply did not come back: labour shortages (80,000 to 100,000 workers, according to the sector49), slow permitting, land costs, taxes (up to 40% of the cost, according to developers49), or existing homes that do not come onto the market. I did not get as far as measuring these factors.
The special study in the June Economic Bulletin partly answers the question left above73:
The same study rules out two popular explanations. Prices and rents rose together (about 88% and 81% per square metre between 2017 and 2024), and consumers' expectations are in line with the actual rise: Banco de Portugal reads this as a market driven by fundamentals, not by a bubble73. And credit followed prices instead of pushing them: the weight of credit in the value of sales stayed below 60%, close to 2019, and the municipalities that rose most were not those that relied most on credit73.
A note on timing: Banco de Portugal estimates that the gap between new homes and new households, about 300,000 over a decade, closed in 2025 as a flow, with less immigration and more permits (41,900 homes permitted, the highest since 2008)73. The accumulated shortfall does not disappear; what stops growing is the hole. If the reading is right, demographic pressure on prices eases from 2026.
Claims that circulate
Claims that circulate about housing and that the data does not confirm.
There are 723,000 empty homes, so there is no shortage.Incomplete
The 723,215 vacant homes in the 2021 Census exist12, but only some of them are for sale or rent and in good condition: 154,075 according to INE and LNEC12, 236,927 according to IHRU93secondary. And many are where nobody is looking for a home.
See chapters 8 and 9 →Foreigners buy 28% of homes.It counts something else
The 28% are buyers of foreign nationality, resident in Portugal or not43. Buyers with tax residence outside the country were 5.0% of sales in 2025 (INE).
See chapters 6 and 9 →Golden visas pushed prices up.Exaggerated
They brought about 2% to 3% of the value of a year's sales62calculation and pushed up the price of homes around 500,000 euros by more than 10%67. Prices kept rising after the property route ended in 2023.
See chapter 6 →It's a bubble and it will burst.Not confirmed
Prices and rents rose together, families' expectations follow the actual rise and credit followed prices instead of pushing them. Banco de Portugal reads this as scarcity, not a bubble73.
See chapter 10 →Short-term rentals took homes away from the Portuguese.True in one neighbourhood, not in the country
In Santa Maria Maior, family homes fell 28% in a decade31, 34. At national scale, short-term rentals are about 2% of homes12, 40calculation, and the cancellation of thousands of registrations in Lisbon did not return homes to housing39.
See chapter 5 →Lisbon and Porto are where supply responds least.A misreading of a study
That is how the press summarised João Amador's study. In the study, Lisbon has the highest supply elasticity in the country and Sintra the smallest effect of demand on prices42.
See chapter 9 →For families
The chapters tell how we got here. This part looks at today's market, in November 2026, and at what a family can do with it.
Test
Choose the region, the size of the home and, if you like, your salary. The test does the sums with the numbers in this analysis. Nothing you type leaves your screen.
Continue
If these topics interest you, leave your email and get the next analyses.
The sign-up form opens here and is run by E-goi. You will then get an email to confirm. Unsubscribe whenever you like; your data is not shared.
Method
BIS quarterly series for Portugal (nominal and real) and Spain (real), read quarter by quarter; annual averages and changes calculated here. Implied inflation is the ratio between the nominal and the real index.
12-month Euribor, average of the twelve months of each year, from Banco de España's monthly table reproduced in4.
INE annual and quarterly reports opened during the research (2013, 2016, 2020 and 2022 editions and Q4 2025), plus news reports citing INE when the report could not be opened.
PORDATA series (INE and SEF/AIMA) from 2000 to 2023 and the AIMA report for 2024.
national register (via the press), two Ministry of the Economy studies, a study for the parish councils of the historic centre, an academic article on Alfama, a Técnico dissertation and the Nova SBE study.
The regional test (new homes for each new household, by region and municipality), the cross-check of prices against the weight of buyers from abroad, the four regional portraits, the years of salary for 90 m² and the table of the weight of each factor in each block are my own synthesis. They rest on public data from INE, Banco de Portugal and the BIS, but they do not exist in this form in any of the sources.
I trained estate agents and property finders in real-estate marketing and sales; they used the training as continuing education to renew the estate agency licence until 2013. I still work with the sector as a consultant, in sales and marketing. No organisation funded this analysis.
Sources
Numbers marked calculation are my own sums on the source given; secondary means I read the number in a news report or another study, not in the original.
Portugal is not short of homes. It is short of homes where people want to live.
Talk to NelsonAnalysis No. 5 · November 2026 · Nelson Viegas · Superior Training Technology