Irregularities in the 2023 Albanian census data

Published in

Busull

8 min readNov 3, 2024

After a long period of delay, INSTAT (Albania’s Institute of Statistics) finally launched the final results of the 2023 census on late June this year. The census had initially been scheduled to have been completed in 2020, but due to the COVID pandemic, as well as the 2021 elections, it was postponed until 2022. Then, it was delayed for another year due to an issue with procuring the tablets necessary for the conduct of the process. The process itself was marred by various difficulties, and the preliminary results were initially supposed to have been launched in December 2023. Thus, we are multiple years behind the initial schedule.

I have been analyzing the detailed results provided by INSTAT, especially after the publication of the regional data, and I have spotted a number of irregularities and suspicious data that have a high probability of not representing the demographic reality. In this article, I will provide my main findings.

The mystery of Beratas in Dibër and Elbasan

When the first results dropped, one of the tables that attracted my attention the most was the one regarding the prefectures of birthplace by prefecture of residence, represented in a matrix.

Country and prefecture of birth for each prefecture of usual residence (doesn’t include all of the latter)

What can be noticed in the table is the abnormally high number of people born in the prefecture of Berat that are residing in the prefecture of Dibër, meaning that nearly 10% of the residents of Dibër (situated in the northeast) come from Berat (located in southern Albania). The same could be said about Elbasan, which finds 11.6k people from Berat. It is highly abnormal because, first of all, it does not conform to any migratory logic during these last 30 years, as people have mostly migrated to the capital, as well as towards the western regions, but also because there is a lot of data contradicting these results.

Checking on data from the civil registry of 2008, we can find in which prefectures those born in the prefecture of Berat were registered in:

Prefecture of registration of those born in the prefecture of Berat in the civil registry of 2008

The data shows a minuscule amount of people born in Berat registered in Dibër, as well as a significantly much lower number registered in Elbasan. We also have data from the 2011 census showing movements from 2001 to 2011 below:

The matrix of internal migration in Albania from 2001 to 2011. 283 individuals moved from Dibër to Berat and 569 individuals moved from Berat to Dibër.

While the data here doesn’t show places of birth, it once again shows how little movement there has been between the prefectures of Berat to those of Dibër and Elbasan, much lower than that implied by the 2023 census.

The most damning evidence, though, comes from the detailed regional results by prefecture. According to the detailed data by municipality, these 10k Beratas are residing almost exclusively in Dibër and Bulqizë, with very few in Mat and Klos:

Then, when we look at the results pertaining the religious and ethnic affiliation of the residents of Dibër, we see a large number of people whose religion and ethnicity is not available (different from those who simply refused to answer), one of the largest ones in proportion to the population of a prefecture in the country:

Almost 16% of the population in Dibër didn’t have any data regarding their religion and ethnicity

It is almost sure that those born in Berat are contained within this category, strengthened by the very low number of declared Orthodox Christians. If the Beratas in Dibër were a somewhat representative sample of the population there, then we should have seen a substantially larger number of Orthodox Christians in Dibër as well.

The results are just as suspicious in Elbasan. While Elbasan itself is a large industrial city where people have moved to from various regions, particularly during the communist period, the detailed data shows that those born in Berat are not actually mostly located in Elbasan, but in two smaller municipalities in Belsh and Cërrik.

According to the census, nearly 25% of the population in Belsh and Cërrik was born in Berat

The data from the 2008 registry shows only around 580 individuals born in the prefecture of Berat registered in the municipality of Belsh, with the real number of residents from Berat being lower due to outmigration since then. In the same way, we can find only around 850 individuals from Berat registered in Cërrik, much lower than the numbers seen above. Once again, the prefecture shows a high number of people without any available data on religion or ethnicity (around 16k), a category which very probably contains these phantom inhabitants from Berat.

Berat’s stable population in the last 12 years

While we have been analyzing the various trajectories of those born in Berat, we also need to look at the prefecture of Berat in itself, because they were the only prefecture outside Tirana that didn’t post any population loss, which is extremely weird to say the least. Once again, we look at the proportion of those for whom we don’t have any data on religion or ethnicity:

Around 26.5% of the population in Berat has no available data, the highest in the country

The overall rate of “no available data” regarding religion and ethnicity is around 5.5% nationally, while the proportion in Berat is practically 5 times larger than the average, at 26.5%. There is no mystery about how Berat has managed to retain the same population for the last 12 years. They simply haven’t, and we can safely assume the numbers here do not represent the reality.

Inflated numbers in the Greek minority areas

The last two suspicious population changes that we can observe relate to the increase in population in the two Greek minority municipalities in the South, respectively Dropull and Finiq. Dropull jumped from a registered population of 3503 in 2011 to 8259 in 2023, more than doubling, while Finiq increased from 10529 to 11413. The jump in Dropull is particularly jarring. It cannot be explained by any population returning from Greece, as the number of births in this municipality during the last 10 years has been extremely low, while deaths have been much greater in number. In Dropull, from 2011 to 2022, there have been only 80 births and 999 deaths. In Finiq, during the same period, there have been 538 births and 1680 deaths. Finiq is in a somewhat better situation, but they are both ultimately in a demographic spiral. Their growth simply hasn’t come from any demographic growth, either from natural increase or migration.

The most plausible alternative explanation is that in 2011 there was a boycott of the census by the Greek minority, which resulted in their undercounting, which could explain to some measure the situation in Dropull. It is hard to measure this, but I believe that the best way we can understand whether the population figures are real or not is to compare them to some external figure that should correlate with the resident population. The best indicator, in my view, should be the turnout in the 2021 parliamentary elections. What makes this an even better experiment, is that these elections were held during pandemic restrictions which meant that only those who actually were residing in the country at the time could vote. This measure can also help us understand other unexpected numbers in the rest of the country. The Greek minority had no reason to not turn out in these elections, being fairly represented in Albanian politics by two parties catering to them (PBDNJ and MEGA), apart from the other major parties.

The overall ratio of the turnout to the population is at 69.2% nationally. Municipalities with a considerably lower ratio can be suspected to have inflated population numbers, with the only plausible explanation that residents there are registered in other places in Albania (mostly true in the capital region). These are the 10 municipalities with the lowest turnout/population ratio:

Notice any pattern? 9 of the 10 municipalities here are ones that we discussed earlier for their suspicious numbers. Dropull and Finiq, the two Greek minority municipalities, lead the list. Dimal, Berat and Kuçovë represent Berat, Dibër and Bulqizë represent Dibër, and Cërrik + Belsh represent Elbasan. Kamëz can be explained by the fact that it is a municipality located in the capital region where a lot of residents are registered elsewhere. At a closer look, the municipalities in Dibër and Elbasan correspond exactly to the ones where we see the phantom populations from Berat. The other municipalities within these prefectures have normal turnout/population ratios, which only makes the pattern clearer. And of course we noticed that Berat had a population figure that suggested that it had not declined during the last 12 years, but the turnout figures do not comply with that idea.

Conclusion

Apart from the known issues regarding the size of the Greek minority, what can explain the reasons behind these numbers in Berat, Elbasan, and Dibër? One plausible explanation could be schemes regarding elections, as it is well known that these prefectures (Elbasan and Dibër more prominently) have had issues regarding vote-buying from local strongmen who work for major political parties. Or perhaps these are just innocent mistakes from the part of INSTAT, but their repetition in multiple prefectures makes that somewhat doubtful. But it’s hard to tell without more details.

Does this mean that we can reject all the results from this census? I think that most of the results presented still remain within the realm of plausibility, but they should be used with caution, particularly in the case of the prefectures that we discussed above. At the very least, when using the numbers, one can exclude those without any available data regarding ethnicity and religious affiliation. This also implies that the overall population is inflated by at least 60k people. It could of course be possible that the population was undercounted elsewhere, but I would rather use an overall population figure of around 2.3–2.35 million people rather than the 2.4 million that INSTAT has provided to us. We’ll need to wait for the publication of further reports and results by INSTAT, including microdata, to further judge on the validity of the rest of the data.