This issue aims to clarify several of Turnleaf’s and the market’s forecast deviations over the past few months. Below, we outline key insights and performance drivers across a selected group of countries. If you have any questions or would like to discuss further, please don’t hesitate to reach out to our team.
Poland
Every year in March, Poland revises its CPI basket to reflect evolving consumer spending patterns. This year, the weights for food, energy, and fuel—components known for their high volatility—were reduced, resulting in revised estimates for January and February that were lower than the preliminary figures. The first notable surprise appeared in January 2025 (TL5.1, MEDIAN BBG 5.0, Realized (Prelim) 5.3, Realized (Revised) 4.9). This revision also affected February, slightly pushing our estimate higher (TL 5.24, MEDIAN BBG 5.3, Realized (Prelim) 4.9, Realized (Revised) 4.7). Since the revisions were applied retroactively, Turnleaf’s February estimate was impacted by these reweighting distortions. Had our model incorporated the updated basket in real time, our February estimate would have been closer to the revised number. Once the volatility settled, Turnleaf’s March estimate aligned more closely with the final outcome and outperformed consensus (TL 4.97, MEDIAN BBG 5.1, Realized 4.9)—as well as effectively capturing the inflationary impact of excise tax hikes on tobacco and alcohol. The inflation outlook for 2025 will largely depend on whether the energy shield is extended beyond September and how energy prices respond to the reinstatement of the capacity fee in June.
India
The January discrepancy was mainly driven by a surprise sharp decline in vegetable prices due to favorable harvest conditions from the previous season (TL 4.57, MEDIAN BBG 4.5, Realized 4.31). For February (TL3.92, MEDIAN BBG 4, Realized 3.61), although our model captured vegetable price fluctuations better than consensus, performance can be improved through the incorporation of city-specific food baskets to account for ad-hoc implementation of food subsidies by the government. For example, Mumbai exhibited steeper price declines relative to the national average, implying that CPI impacts may be underrepresented, especially in regions with higher index weights.
South Korea
From January (TL 1.99, MEDIAN BBG 2.1, Realized 2.24) to February 2025 (TL2.12, MEDIAN BBG 2.1, Realized 2.02), the inflation landscape was characterized by high currency volatility and distortions in the energy market. Our model performed in line with consensus indicating that it was a broad surprise for many. Going forward, we are focusing on improving estimates of currency pass-through, particularly considering upcoming shifts in energy policy: the reintroduction of LNG tariffs in April, the return of consumption taxes on LNG and coal in July, and tuition fee hikes in September.
South Africa
Food inflation slowed significantly, though limited access to granular food price data weakened model sensitivity to these changes for January (TL 3.24, MEDIANBBG 3.2, Realized 3.19) and February (TL 3.36, MEDIAN BBG 3.4, Realized 3.16).Additionally, changes in CPI weights as of January appear to have distorted seasonal patterns for February, yet Turnleaf performed slightly better than short-terms. Key inflation risks ahead include the VAT increase in May and mid-year hikes in electricity tariffs. Given South Africa’s reliance on global energy imports, exchange rate dynamics will be crucial throughout 2025.
Turkey
Turnleaf’s forecasts have remained above both Bloomberg and the central bank’s projections toward the tail end of the forecast horizon. Although we accounted for the minimum wage and fuel duty adjustments, our January forecast (TL 41.02, MEDIAN BBG 41.1, Realized 42.12) did not have sufficient data to adequately capture the broad increases in healthcare services. Again, for February, data limitations in capturing volatile healthcare prices following government intervention to reduce co-payments at private and public hospitals kept TL’s estimates slightly away from the realized figure (TL 39.91, MEDIAN BBG 39.9, Realized 39.05). A sharp depreciation in the lira, driven by political unrest, further complicated the inflation trajectory. Future outcomes will depend on macroprudential policy choices and liquidity measures from the central bank.
It is important to emphasize that the beginning of the year is often difficult to predict given sudden changes in service prices that may carry lagged price effects towards subsequent months through sudden government intervention and spillover effects to other industries.
Czech Republic
In January, a rise in food and beverage inflation pushed prices higher than traditional repricing expected by the central bank forecasts at the start of the year—leading to a broad surprise for everyone (TL 2.55, MEDIAN BBG 2.6, Realized 2.8). Particularly, the increase was mainly driven by spirits (10.8%), fruits (6.5%), and vegetables (7.1%) for which our data only contained a subset with the highest basket weightings (e.g. potatoes and apples) that didn’t fully capture this swing. In February, retail energy prices, which is underrepresented in our data(especially electricity and gas) continued to fall as suppliers competed to regain market share post-crisis (TL 2.8, MEDIAN BBG 2.7, Realized 2.7). A continued downside risk remains, as local sources report further cuts in retail energy prices.
Hungary
A substantial surprise emerged in January service fee inflation, which our model failed to capture due to data limitations (TL 4.77, MEDIAN BBG 4.8,Realized 5.5). In contrast, February food inflation—also a notable shock—was mostly anticipated (TL 5.32, MEDIAN BBG 5.3, Realized 5.57) with lagged price effects from January likely carried over to February 2025. The inflation surprise led the Orban government to impose profit margin caps on a list of 30essential food items. However, the policy impact remains ambiguous: while intended to mitigate inflation, these controls may amplify price pressures elsewhere as shortages emerge and consumers shift demand toward non-controlled goods.