Featured Graduate Student Research: Jonas Mueller-Gastell

Historical Background

Before World War 2 had become a global conflict, Finland found itself invaded by the armies of the Soviet Union. Soon it had to cede much of the historical region of Karelia, with its industrial heart Vyborg. To protect the population in the ceded territory, Finland evacuated all of the approximately 400,000 inhabitants of Karelia (more than 10% of Finland’s total population) and resettled them in the remaining country. After Germany’s invasion of the Soviet Union in 1941, Finland tried to retake its lost territory, but – after occupying it for three years and seeing most of the former population return – in 1944 had to sign a separate armistice with the Soviet Union again. It now lost Karelia a second time, in addition to some further northern territories and most of the islands to its southern shores, and had imposed upon it heavy financial reparation demands. Finland hence had to perform the impressive feat of evacuating and finding shelter for 440,000 people once more.

Finland was a late comer to the industrial revolution, with most of the country still agricultural in the 1930s, a large share of subsistence economy, and industrial production concentrated in timber products and paper mills. But, in part due to the administrative legacy of Swedish rule, it also had an efficiently run public administration and efficacious government services: the task of resettling the displaced, expropriating large land-owners (to find space for farms for Karelian peasants), and taxing capital (to fund reimbursements to Karelian property holders in general and to pay reparations) was completed quickly. By 1948, all Karelians had found a permanent home and were integrating economically with their new host communities. A side effect of this bureaucracy: detailed data was collected on all stages of the taxation, expropriation, compensation, and resettlement program, in addition to detailed census data collected in 1950 and thereafter.

Project Description

The unique combination of large scale resettlement, an industrializing economy, a large government intervention in the economy in the form of capital seizures and land reform, and detailed data collection poses a fantastic opportunity to test many theories in the social sciences. I was particularly interested in two sets of questions: 1) How quickly do Karelians (and other resettled Finns) integrate or “assimilate” in their host communities? Finland has high ethnic homogeneity (e.g., Karelians and other Finns are mostly Lutheran christian and speak the same language, albeit with minor regional dialects). The experience of the Karelian Finns hence gives us some sense of a “lower bound” on the difficulty of integrating a large number of newcomers in a community. Understanding the behavioral patterns that do and do not differ after several years or even generations between resettled and natives may inform policy making in the wake of the current age of mass migration. 2) When land and capital are strongly redistributed, how long does it take for wealth inequality to reach previous levels (if at all), and how does investment and savings behavior respond in the short and medium rum? Land reform is a topic studied often in instances of low data availability and usually few long-term outcomes can be observed. In contrast, the 70 year multi-generational time-frame of the Finnish case allows for detailed study of how redistribution percolates through generations.

Purpose of Trip to Finland

Both of these questions can be answered by using the excellent Statistics Finland data infrastructure (which includes 10% of the 1950 Census, 100% of the 1970, 1975, 1980, 1985 Censuses, and large amounts of yearly registry data from 1987 till today) in conjunction with historical archive records. However, Statistics Finland restricts access to census data products tho those physically present in Finland (or the EU). Similarly, most of the detailed archival data has not yet been scanned for digitization, and hence requires trips to Finland.

In my first three week trip to Finland, funded by the Europe Center grant, I decided to concentrate on acquainting myself with the data infrastructure, gathering information on data quality and reliability, running exploratory analyses on the questions posed, and establishing connections with the researchers working on topics of Finnish history and labor markets, at Aalto University (where I had graciously been given an office to work from) and other universities. I will return again in September to continue my Census-based data work and push forward several digitization efforts, in collaboration with Finnish researchers.

Preliminary Results on Assimilation

As a first measure of how long the identity of being resettled or Karelian affects important life decision, I decided to study assortative family formation. More precisely, I wanted to find out for how long the category “resettled/ Karelian” is a relevant one for who decides to have children with whom.1 Statistics Finland provided me with the complete set of all children born from 1970 onwards and the (encrypted) ID code of their parents, as well as a large percentage of all children born before 1970 plus encrypted parental ID codes.

The 10% digitized extract from the 1950 Census contains several retrospective questions to 1939, i.e., pre-resettlement. I can hence code those who lived in the ceded territories in 1939 as “resettled” while everyone else is coded as “native.” Then, I link the reshaped data-set of births to the 1950 Census extract, thereby identifying who has a child with whom. In all analyses, I restrict attention to the first child for any given couple (not necessarily the first child for any one member of the couple, however). An alternative research design utilizes the much larger data set of the 1970 Census (where 100% is available): using the municipality of birth, I can delineate people into “Karelian” and “not Karelian”.2 I perform two main pieces of analysis for the two resulting data sets: 1) for any given year, I regress a dummy for “has a child with a Karelian/resettled person in this year” on an indicator for being resettled/Karelian oneself, the percentage of Karelian/resettled people of total municipality population (of municipality of residence in 1950/1970 respectively), several control variables such as a fourth order polynomial in age, and a gender dummy; 2) instead of controlling for a host of covariates, I subset the data set to those who in fact have a child in the given year and then regress the Karelian/resettled partner dummy on the Karelian/resettled dummy and the percentage of Karelians/resettled in the municipality. The coefficient of interest, in either specification is the coefficient on the Karelian dummy. This coefficient tells us how much more likely a Karelian person is to form a family with another Karelian person, relative to the average non-Karelian Finn. Figure 1 plots the coefficient results (in percentage points) for the smaller 1950 Census based data set, Figure 2 plots the coefficients for the larger 1970 Census based data set, with 95% confidence intervals.

We can see that results are qualitatively quite similar across the two analyses types and the two data sources. In 1944, Karelians/resettled people were much more likely to partner with another resettled person (potentially reflecting prior contact and ties), with the effect roughly halving by 1950. However, the effect does not disappear immediately after original ties could be expected to have been exhausted. In the (larger and thus more statistically powerful) 1970 Census based data set, we can detect significant positive effects of being resettled up until the 1970s. Thus, for more than 20 years, being from Karelia had a sizable and significant effect on an extremely important life choice: who to have children with.

Future Plans

I have also explored several other avenues of quantifying and measuring assimilation or integration. In fact, there are significant long-run effects of being from Karelia (or having a mother or father from Karelia) on many economically relevant variables: e.g., we can detect a (positive!) effect of resettlement on income in 1970, 1975, 1980, and even 1985 for first generation and second generation resettled people. I chose to focus on the results for family formation in this report, as they are both the most novel and neatly illustrative, I believe (and I do not have full clearance from Statistics Finland on all of my other results yet).

In future trips to Finland, I am looking forward to pushing ahead on more traditionally economic measures of, e.g., labor market integration and assimilation, as well as on the topic of wealth inequality and land reform. Thanks to the generous funding from the Europe Center, I was able to make much progress on my endeavor to learn from the unique event of the Karelian resettlement, and have made many valuable connections with researchers in Finland.

_____________________

1A complementary analysis would look at marriage or household formation in general, rather than child-bearing. I am currently applying to be given access to the data-set of all marriages formed from 1950 onwards, as well as the data fields that would allow me to identify (married or non-married) household units in 1970.

2Of course, one needs to note that the municipality of birth does not guarantee being a long-term resident and thus may mis-identify those who moved early in life to Karelia or those who left early I perform some checks on the set of individuals that I can match between the 1950 Census extract and the full 1970 Census. Approximately 80% of the classifications match, i.e., are identified as Karelians/resettled in either the 1939 definition or the birth municipality definition.

Jonas Mueller-Gastell is a PhD candidate in Economics.