Mandatory pension funds are the largest domestic institutional investors in Croatia, managing EUR 28.7bn on behalf of more than 2.4 million members as of 30 June 2026. Despite that scale, the decisions that shape how most members’ savings are invested are still driven largely by statutory rules and automatic allocation. This gives us a good reason to look at the second pillar from the inside: who participates in it, how the assets accumulated, what the funds earned and how much the choice of category and pension company ultimately mattered. One caveat is necessary before looking at the numbers: the period analysed here, from the end of 2020 to June 2026, is much shorter than the investment horizon of a typical pension saver and should not be interpreted as a forecast of future returns.

Croatia’s pension system has operated in its current three-pillar form since 1 January 2002. The first pillar is the pay-as-you-go system administered by HZMO and financed through current pension contributions and transfers from the state budget. The second pillar is a mandatory defined-contribution system based on individual accounts: for members insured in both mandatory pillars, 15% of the relevant gross wage base is paid into the first pillar and 5% into the member’s personal account in a mandatory pension fund. The third pillar consists of voluntary retirement savings.

Four pension companies operate Croatia’s mandatory pension funds, and each offers categories A, B and C. Category B is the legal successor to the original funds launched in April 2002, while categories A and C were introduced in August 2014. Category A allows the highest equity exposure, category B follows a more balanced policy, and category C has the most conservative mandate. The maximum permitted equity exposure is 70% for A, 40% for B and 10% for C, while the minimum allocation to qualifying public-debt securities and money-market instruments currently moves in the opposite direction, at 25%, 45% and 60%, respectively. These limits create different risk profiles, but they still leave pension companies considerable room to make their own investment decisions.

Cumulative unit values and MIREX, indexed to 100 on 31 December 2020

Source: Pension fund management companies, HANFA, REGOS, InterCapital Research

Starting from a value of 100 at the end of 2020, MIREX A, the net-asset-weighted benchmark for category A funds, reached 185.0 on 30 June 2026, while MIREX B reached 142.9. This translates into cumulative nominal returns of 85.0% for category A and 42.9% for category B, equivalent to annualised returns of approximately 11.8% and 6.7%, respectively. Three of the four A funds finished above every B fund, although the weakest A fund underperformed the two strongest B funds. Category was therefore a major determinant of performance, but it was clearly not the only one.

The difference between MIREX A and MIREX B amounted to 42.1 percentage points. Within category A, however, cumulative returns ranged from 44.3% to 94.2%, creating an even wider spread of 49.8 percentage points. Category B returns ranged from 26.6% to 63.2%, a spread of 36.6 percentage points. Funds operating under the same category-level limits therefore produced very different outcomes, reflecting differences in asset allocation, security selection, bond duration, currency exposure and the timing of portfolio decisions.

On 30 June 2026, the mandatory funds had 2,413,050 members: 558,804 in category A, 1,779,254 in B and 74,992 in C. Membership should not be equated directly with the number of active contributors, but the scale can be compared with the preliminary HZMO figure of 1,798,712 insured persons at the same reporting date. The more interesting development is the shift between categories. Category A grew from 150,266 members at the end of 2021 to 558,804 by June 2026, an increase of approximately 3.7 times, while category B declined from 1,899,623 to 1,779,254.

Members by category (Dec 2021–Jun 2026, thousand members)

Source: HANFA, REGOS, InterCapital Research

Part of this shift reflects members actively changing categories, but automatic allocation has also played an important role. Since October 2019, new entrants who do not select a fund within the statutory deadline have generally been assigned by REGOS to category A. Amendments adopted in 2023 extended the period during which automatically allocated members may remain in A from ten to fifteen years, although this change cannot yet explain the growth recorded through June 2026. The first cohort covered by the October 2019 rule would have reached the previous ten-year limit only in 2029.

Automatic allocation remains dominant. In 2025, REGOS assigned 85,270 of 89,068 new members ex officio, equivalent to 95.7%, while only 3,798 selected a fund themselves. There are, however, early signs of greater activity. During the first half of 2026, 7.1% of new entrants selected a fund themselves, compared with 2.7% in the same period of 2025. By the end of May, 15,100 existing members had also changed funds at their own request, compared with 7,981 during the whole of 2025. The data do not reveal whether this was driven by performance differences, marketing or greater public interest, but the change is large enough to follow.

Turning to the money, net assets reached EUR 28.73bn on 30 June 2026, of which category B held EUR 25.44bn, category A EUR 1.77bn and category C EUR 1.51bn. Category A therefore accounted for only 6.2% of assets despite representing more than 23% of members, mainly because its members are generally younger and have had less time to accumulate savings. At the end of 2025, total mandatory-fund assets represented approximately 28.5% of Croatia’s GDP.

OECD data offer a useful regional comparison, although their pension-provider measure is broader than Croatia’s mandatory second pillar alone. At the end of 2024, Croatian pension-provider assets amounted to 29.6% of GDP, compared with 16.2% in Slovakia, 8.9% in Romania, 8.4% in Poland and 4.8% in Hungary. Croatia therefore recorded the highest ratio among these selected Central and Eastern European countries. The OECD figure and the mandatory-fund figure are not perfectly interchangeable, but both show the importance of funded pension assets within Croatia’s financial system.

Net assets by category (Dec 2011 – Jun 2026, EURbn)

Source: HANFA, InterCapital Research

From the end of 2021 to June 2026, category B assets increased from EUR 16.3bn to EUR 25.4bn, a rise of approximately 56%, while category A expanded from EUR 0.19bn to EUR 1.77bn. The only year-on-year decline in total net assets visible in the complete archived year-end series occurred in 2022. Despite a net inflow of approximately EUR 812m, assets fell by EUR 107m because the estimated investment result was a loss of roughly EUR 919m. The second pillar continued receiving contributions, but falling market values more than offset the new money entering the system.

An increase in assets does not necessarily represent investment performance. Assets grow through contributions, decline through payouts and fluctuate with market prices. At the level of the whole system, transfers between funds cancel out, allowing the annual change in assets to be separated into net inflows and a residual investment result. In 2024, for example, assets increased by EUR 2.99bn, of which approximately EUR 1.12bn came from net inflows and EUR 1.87bn, or 62.6%, from investment performance. The investment-result share reached 70.7% in the first half of 2026.

Decomposition of the change in net assets (2021–H1 2026, EURm)

Source: HANFA, InterCapital Research

The cumulative picture is more balanced. From the launch of the system to 30 June 2026, members paid approximately EUR 19.04bn in contributions, while EUR 3.42bn was paid out from closed accounts. Cumulative contributions net of payouts therefore amounted to EUR 15.62bn. With current assets of EUR 28.73bn, the residual attributable to investment performance stands at approximately EUR 13.11bn. On this basis, investment performance accounts for 45.6% of current assets, while contributions net of payouts account for 54.4%. Almost half of today’s second-pillar assets were generated by accumulated returns, but contributions remain the larger cumulative source of the asset stock.

The regular contribution flow also gives the funds a stable source of investable cash. Because it is linked mainly to employment and wages, this inflow continued during the 2022 market decline and allowed funds to purchase securities at lower prices. Market gains increase the value of existing portfolios but do not themselves create cash for further purchases. That funding comes from new contributions and investment income such as coupons and dividends.

Annual and cumulative returns by fund and category (2021- H1 2026, %)

Source: Pension fund management companies, HANFA, Eurostat, InterCapital Research

Fund performance is measured through changes in the value of the accounting unit, which already reflects costs charged directly to fund assets, including management fees. It does not include every possible cost borne by an individual member, such as an applicable exit fee, and a member’s personal result also depends on when contributions were paid. The figures should therefore be understood as fund-level time-weighted returns rather than returns earned identically by every member.

Category A outperformed category B at the same pension company in 18 of 20 full-calendar-year comparisons between 2021 and 2025 and in all four comparisons during the first half of 2026, bringing the total to 22 out of 24 comparable periods. It also delivered the highest cumulative return at all four companies. This is consistent with its higher equity exposure during a period in which equities, including Croatian equities, performed strongly. It should not be read as evidence that category A will outperform in every environment, since the same mandate also exposes members to larger potential equity-market losses.

The year 2022 showed why the relationship between the statutory category and realised risk is not always straightforward. MIREX A declined by 4.1%, while MIREX B fell by 5.0%, and at three of the four pension companies the A fund lost less than the corresponding B fund. CROBIS fell by 12.6% and CROBEX by 4.8% that year, meaning that category B’s larger bond allocation offered little protection during an unusually severe bond-market sell-off.

Inflation provides the most relevant benchmark for pension savings. Croatia’s Harmonised Index of Consumer Prices increased by 35.4% between December 2020 and December 2025. Over the same five calendar years, MIREX A returned 64.3%, while MIREX B returned 34.1%. After adjusting geometrically for inflation, MIREX A generated a cumulative real return of approximately 21.3%, while MIREX B recorded approximately –1.0%, leaving its purchasing power broadly unchanged at the end of 2025.

All four category A funds finished the five-year period with positive cumulative real returns, ranging from approximately 0.5% to 29.3%. Category B outcomes were mixed, ranging from approximately –9.4% to +9.3%. Results in the first half of 2026 improved the cumulative position and, on an estimated linked HICP basis, brought MIREX B’s real return from the end of 2020 to June 2026 to approximately 2.2%.

Putting everything together, the second pillar has grown into something the domestic market cannot ignore. Its EUR 28.7bn of assets is equivalent to roughly 28.5% of GDP, against a total market capitalisation of all shares listed on the Zagreb Stock Exchange of about 34.6% of GDP at the end of 2025. In other words, one group of four managers now runs a pool of savings not far off the value of the entire domestic equity market, and that pool keeps growing every month regardless of what prices do, because contributions follow wages rather than markets. That is a good problem to have, but it is still a problem: in a market where the number of listed shares has fallen from 147 at the end of 2016 to 77 at the end of 2025, a permanent buyer of this size runs into the limits of what is available to buy. The constraint is not the funds’ willingness to invest at home; it is the supply of investable domestic paper. More listings, more free float and more frequent capital raises would allow a larger share of this money to stay and work in Croatia, and would make the market deeper for everyone else trading in it.

The second shift is happening on the members’ side. Category B is slowly shrinking while category A is growing fast, and most of that is legislated rather than chosen, since anyone who does not pick a fund is placed in A by default. But 2026 brought the first real signs of members acting on their own, and the returns above show why it matters: over the period analysed, the strongest funds delivered roughly twice the cumulative return of the weakest within the same category and the same legal limits. A member in their twenties has decades ahead of them, and for that horizon equity exposure is the normal setting rather than the risky one. The next post turns from members to the portfolios themselves: how much of this money sits on the Zagreb Stock Exchange, in which companies, and what ownership of that size does to liquidity.