About the population

The population data is properly documented by the INE from 1964 onwards, but in order to approximate the population growth between 1900 and 1963, the reference data used has been the historical series provided by the INE (Instituto Nacional de Estadística). Link source:

• Series históricas 1900-2001: Población por sexo y edad - Datos censales.

Since the data provided by the INE only corresponds to data from the beginning of each decade (1900, 1910, 1920...), data from the work from Roser Nicolau has also been used to approximate the population for each of the years between 1900 and 1963.

The data presented by Roser Nicolau uses the mentioned historical data by the INE, but does a deeper sourcing work in trying to provide better accuracy on the Spanish demography. Link source:

• "Estadisticas Historicas de España, siglo XIX - XX" (2005, second edition), chapter 2: Población, salud y actividad, by Roser Nicolau, Page 124, Cuadro 2.5. Book coordinated by Albert Carreras and Xavier Tafunell, Fundación BBVA, ISBN: 84‑96515‑00‑1.

The next table shows a comparison of the data from the INE, Roser Nicolau estimations and our own estimations:

We respected the data provided by the INE for the beginning of each decade, with the only difference existing in the year 1940, were we reduced the estimated population in comparison to the one provided by the INE, and also being less than Nicolau's estimates.

The reason for downgrading the 1940 data is related to the Spanish Civil War. Although most estimates prefer not to infer population loss caused by the war, even official sources of the time assume a population loss at least in 1939:

• "Movimiento Natural de la Población de España, año 1939"(1943), done by Talleres del Instituto Geográfico y Catastral, Ministerio de Trabajo, Direccion General de Estadistica, Madrid.

We do not venture to assume any sort of population loss either, even if it is highly probable at least in 1939 (even more so if we consider migratory movements), but what we do consider reasonable referring to the "Movimiento Natural de la Población de España, año 1939", is that in any case population growth suffered a fall compared to previous years in 1937, 1938 and 1939.

About the housing stock

The Spanish Government has adequately documented housing stock data from 2001 onwards. However, to approximate housing growth between 1900 and 2000, we relied on several sources.

From 1970 to 2000, we utilized data sourced from the Atlas Digital de las Áreas Urbanas (Parque de Viviendas > Viviendas según tipo > Viviendas familiares > Select from available years 1970, 1981, 1991, 2001, 2011 > Tabla and Sintesis): Link. Its database provides housing stock figures for the years 1970, 1981, 1991 and 2001, with data sourced from the INE (Censo de Población y Vivienda) according to its notes.

However, data for the years in between (1971, 1972...1980...1994...2000) are not available. To fill these gaps, we relied on estimates provided by the Banco de España, which covers the period from 1981 to 2000 (Observatorio de Vivienda y Suelo > Boletines periódicos > Boletín Anual 2021 > Tabla 8.1): Link. We made slight adjustments to ensure consistency across our comprehensive dataset. For comparison, here are some sample figures:

From 1900 to 1970 we have made our own estimations based in data collected by Xavier Tafunell (2005) in:

• "Estadisticas Historicas de España, siglo XIX - XX" (2005, second edition), chapter 6: Urbanización y vivienda, by Xavier Tafunell, Page 491, Cuadro 6.7 and Cuadro 6.6. Book coordinated by Albert Carreras and Xavier Tafunell, Fundación BBVA, ISBN: 84‑96515‑00‑1.

From 1950 until 1969, we replicated and used the data from Tafunell (2005, page 491, table 6.7), incorporating minor adjustments to ensure consistency across our comprehensive dataset. As means of comparison, here there is a sample from Xavier Tafunell and our own estimation:

A discrepancy in the sample can be observed between 1900 and 1949. This is because, prior to 1950, Tafunell research only reports housing stock from urban areas, not for rural areas (page 490, table 6.6, "Parque estimado de viviendas en núcleos urbanos por provincias, 1860 - 1950").

The lack of reliable sources for rural areas is unfortunate, as we consider housing in rural areas to be quite relevant for obtaining a comprehensive historical overview. If we accept the figure of 1,879,400 houses in 1900 as accurate, the annual construction estimates from 1900 to 1949 would yield improbable values given the resources and means of production available during that period. To address this data gap, we had to walk on thin ice.

According to research conducted by the Centre d'Estudis Demogràfics (2), the number of households in the year 1900 is estimated to be around 4.8 million.

For our estimations, we assumed that the actual housing stock in 1900 should have probably been somewhere in between 1.8 million, being the urban housing stock reported by Tafunell the minimum, and 4.8 million, being the number of households reported by Centre d'Estudis Demogràfics the maximum.

Further accuracy is needed since the gap between those two values is significant. To enhance accuracy, we sought guidance from the buildings per period of construction dataset (Viviendas principales por zona y tamaño del municipio, comunidades autónomas y provincias y periodo de construcción del edificio (antes de 1900 a 1950)): Link

Based on the period of construction dataset, between 1900 and 1920, 61% of houses were in urban areas (prior to 1900 would be 52%), while 39% were in rural areas (prior to 1900 would be 48%). Calculating from these figures, if there were 1.8 million urban houses in 1900, which accounted for 61% of the total housing stock, the total number of houses would be approximately 2.9 million. This estimate likely errs on the higher side, as the proportion of rural houses in 1900 was greater than in 1920 (as it was prior to 1900, as data suggests). For this reason, our estimate for 1900 would be somewhere around 3.1 million houses.

Then, having ≈ 3.1 million houses in 1900 as our starting point, one of the last variables we considered for our estimations was the number of people employed in the construction sector, from the historical research conducted by Leandro Prados de la Escosura (2017), found in:

• Spanish Economic Growth, 1850-2015, London: Palgrave Macmillan, spreadsheets and extensive data available in frdelpino.es.

We utilized the number of people employed in the construction sector (Leandro Prados de la Escosura,2017) as a proxy to estimate the number of houses that could have been built each year within the considered range, using the isolated data points collected from several sources as anchors (Atlas Digital de las Áreas Urbanas, Banco de España, Xavier Tafunell and Centre d'Estudis Demogràfics, including our base estimation for the year 1900).

The main question we faced was determining the labour input per unit of output. We have reliable governmental data from 1990 onwards, but the variance is substantial, ranging from 3.3 workers per house constructed in the year 2000 to 14 workers per house constructed in 2020, making accurate estimation challenging.

The best proxy we identified is linked to demography, particularly the number of inhabitants per house. As shown in the previous table, the number of inhabitants per house has been steadily decreasing over the last 50 years (source: INE). It is reasonable to assume that the number of inhabitants per home would have been higher the further back in time we go. In fact, following our assumption of around 3.1 million houses in 1900, together with 18 million inhabitants on that same year (sourced by the INE), the number of inhabitants per house in 1900 would be ≈ 6.

This decreasing trend in the number of inhabitants per house somehow mirrors the trend observed in the total fertility rate, which has also steadily declined from 1900 to the present day. It is reasonable to hypothesize that a reduction in the fertility rate is likely to be associated with a decrease in the number of inhabitants per house. You can review Spain's total fertility rate data here:

• "Estadisticas Historicas de España, siglo XIX - XX" (2005, second edition), chapter 2: Población, salud y actividad, by Roser Nicolau, Page 83, Cuadro 2.2. Book coordinated by Albert Carreras and Xavier Tafunell, Fundación BBVA, ISBN: 84‑96515‑00‑1

In addition to the correlation with the total fertility rate, examining household size may offer further insights. It is important to distinguish between household size and the number of inhabitants per house, as they represent different metrics. Inhabitants per house is calculated by dividing the population by the number of houses, while household size is more closely related to familial relationships, allowing for the possibility of multiple households within a single house.

According to the Centre d'Estudis Demogràfics (2) and Fernando Mikelarena Peña the average household size was around 4.8 people per household in the 1900s.

Our estimations suggest around 6 inhabitants per house in 1900, while the literature indicates a household size of approximately 4.8. This discrepancy could be attributed to the housing shortage at the time, driven by lower productivity and generally poorer economic conditions, which forced multiple households or families to live under the same roof. In contrast, due to the significant construction boom between the 1960s and 2000s, the number of inhabitants per house in 2020 (1.83) is lower than the household size in 2020 (2.51). However, this is not always the case, as the ratio between these two variables changes depending on periods of housing abundance or scarcity relative to the population.

Recently, CaixaBank Research indicated that between 2020 and 2022, around 420,000 new households were created in Spain, while the number of new homes started barely reached 300,000, highlighting a current scenario of scarcity. For this reason, we believe the number of inhabitants per home in the 1900s was likely higher than the average household size of 4.8 people.

Overall, the final estimations suggest an initial requirement of ≈ 8 workers needed to produce 1 house in 1900, similar to the ratio observed in 2010. This would gradually decrease until reaching the well-sourced figure of 3.9 workers per house in the year 1990, following the historical minimum of 2000 during the Spanish housing bubble. After the burst of the bubble, productivity would have plummeted, resulting in the current ratio of 14 workers per house constructed.

Selecting an input per unit of output for the entire period was challenging, and it's essential to proceed with caution when interpreting these numbers. Further research is encouraged to refine these estimates. However, by analysing multiple variables, we believe the estimations provide a realistic depiction of the evolution of the housing stock.

As always, you can check the data on github:

• Download full dataset: Github explorer