Measured against the dividend paid, earnings per unit came to 101% in 2012, 103% in 2013, 100% in 2014, 103% in 2015. During the good years, then, the payout was covered, and retained earnings built back up year after year, from €0.09 per unit to €0.69. Then came 2016: €9.21 of earnings for €9.80 paid out, 94%. Retained earnings fell back from 0.69 to €0.14 per unit17. It was the first year the SCPI paid out more than it earned, and the top-up came out of retained earnings, themselves fed by inflows. The 5% that brought the money in was now, in part, paid for by the money coming in.
Earnings per unit, measured against the dividend paid (axis starting at 85%)
Above the line, the SCPI earns at least what it pays out. In 2016, for the first time, it pays out more than it earns.
Primopierre annual reports 2012 to 2016.
The same year carries the other signal. The financial occupancy rate fell from 91.7% to 82.7% at the end of 2016. Nearly €1 of potential rent in every 5 wasn't coming in. Grégory Frapet, then chief executive of Primonial REIM, offered his justification18:
Primopierre's vacancy, he explained, notably included the temporary vacancy of the Atlantis building in Massy and that of several properties not fully let when they were acquired. So, in his words, it was not structural vacancy.
Grégory Frapet, 2016 annual report
Structural.
Financial occupancy rate, 2009-2025, and what the reports say about it
Grégory Frapet, 2016 annual report:
so it is not structural vacancy.
2017 annual report:
the Atlantis has been re-let, described as the largest vacancy. In the table of vacant space: Between, Golf Park, West Plaza.
Stéphanie Lacroix, 2018 annual report:
the vacancies do not amount to structural vacancy.
2019 annual report:
the main vacancy concerns Golf Park in Toulouse and the Between building. The word “structural” has disappeared.
Financial occupancy rate at year-end. The numbered markers place the published explanations on their year.
Primopierre annual reports 2009 to 2025.
See all the KPIs on the Zero Bullshit website (in French)
The explanation is true, and that's exactly the problem. Every year, the vacancy comes from the most recently bought buildings: not yet let, being marketed, waiting for a tenant. Except that in a machine investing 300 then €400M a year, the most recently bought buildings aren't an anomaly in the portfolio. At some point, the most recently bought buildings become the portfolio.
The following year confirms it, in its own way. The unit price rose from 196 to €208 on 1 April 2017. The temporary problem, for its part, was settled by the book: the Atlantis was re-let in April, and the annual report celebrated19:
The average financial occupancy rate over 2017 came to 88.9%, the report said, rising significantly after the Atlantis in Massy, which had been the largest vacancy, was effectively re-let in April.
2017 annual report
88.9% as an annual average, 91.1% as a year-end snapshot: settling the problem didn't bring the machine back to its earlier levels. And the same report, a few pages later, sets out the table of vacant space for the year: Between, Golf Park, West Plaza19. Then 90.5% in 201820. In the meantime, the management had changed. But not the talking points. Stéphanie Lacroix, the new chief executive, rolled them out20:
The main vacancy, she wrote, concerned Golf Park in Toulouse, but was due to the delivery of an extra 9,000 m² still to be let. The other main financial vacancies (Between, at the foot of the Esplanade de la Défense, and West Plaza in Colombes, in the Hauts-de-Seine, west of Paris) came, she said, from well-located buildings and did not amount to structural vacancy.
Stéphanie Lacroix, 2018 annual report
Structural.
Le Between. Bought for €169M in 2016, on the vacancy table from 2017, there again in 2018. The question ends up asking itself: so what is structural vacancy, then?
Beneath the occupancy rates, the debt is at work. In 2017, €620M was invested, including Cœur Défense and In & Out in Boulogne, through SCIs, property-holding companies. A problem already raised here and here (both in French), and one the regulator should clearly ban. Debt climbed to €558M, or 29.2% of the portfolio, for a ceiling of 30%19. Leverage had become the real engine. 2018 slowed down on the surface: €194M invested, with the W building in Nanterre as the star of the show, €84.7M, let to Faurecia, and a first in the SCPI's history: the sale of about 59% of the units of the SCI Energie+ for €47.5M, with a €15.3M capital gain to boot20. Bingo.
But 2019 rolled on. A record year: €512.8M of subscriptions, which made up for the few withdrawals. The occupancy rate, for its part, fell back to 86.1% at year-end. The annual report recycled the usual reading grid. But the word “structural” had disappeared. That said, the reading grid was the same21:
The main vacancy, according to the report, concerned Golf Park in Toulouse and the Between building at the foot of the Esplanade de la Défense.
2019 annual report
4 years after it was bought, Le Between was still the portfolio's main vacancy. Except that other acquisitions were making the numbers worse.
As for the balance sheet, it treated itself to a reassuring line: debt fell from 542 to €318M. Or not. Because the real deleveraging came from a single line: the SCI Grand Seine sold Le Millénium, Natixis's building, for €251M, and passed up €34.3M in dividends21.
4 years of vacancy explained, never owned. The word that keeps coming back in the reports isn't “temporary”, used only once, for the Atlantis: it's its negation. “Not structural”, written in 2016, written again in 2018, while the list of empty buildings renewed itself underneath. Published, signed, and nobody batting an eyelid. The members first: the product's profile is well known, older, not very mobile holders, and that's not an insult. Then the advisers, who kept repeating that 4.5% was still better than a Livret A savings account at 0.5% that doesn't beat 2% inflation, which is accurate but dubious.
The adviser's arithmetic compares levels, never slopes. While the Livret A served as the bogeyman, the average SCPI yield was melting, decade after decade: between 5.5 and 6.5% in the 2000s, around 5% after the financial crisis, then a steady slide down to 4.4% in 2019. Primopierre followed the same slope, almost to the decimal point: 5.46% in 2011, 4.4% in 2019, excluding distributions of capital gains. That 4.5% was never a risk premium, it was a yield squeezed by free money, delivered by buildings bought at ever higher prices. Over the period, risk was rising, vacancy was settling in, leverage was climbing, and the reward for risk was falling. The saver was buying the gap with the Livret A. What they got was a risk premium in reverse.
The fact remains that between the moment when nothing is structural and the moment when it becomes so, perhaps at the end of 2019, more than €1.1bn of subscriptions came in. The best sales year in the SCPI's history is the one that followed three years of published explanations.
That said, a large part of the inflows came through another pipe: insurers, who hold the units inside their life insurance policies and take their fees on assets that every subscription swells. Suravenir (Crédit Mutuel Arkéa) lists Primopierre in its Sérénipierre policy. The same Suravenir that, incidentally, sits on the SCPI's supervisory board20. While its parent company is a shareholder of Primonial. And while the insurer sits alongside… another Primonial vehicle.
Payout per unit, 2009-2025
Dividend paid for a unit held over the full year. The first financial year ran from August 2008 to December 2009, i.e. 17 months, hence the €17.81 for 2009.
Primopierre annual reports 2009 to 2025.
See all the KPIs on the Zero Bullshit website (in French)
But while the reports and its sellers piled up the explanations, the machine was gearing up for its biggest years. And, incidentally, its worst ones too.