J.P. Morgan31 July 2026

Global Data Watch

And isn’t it ironic?

July central bank meetings reinforced our view that a broad DM hiking cycle is on the horizon. While we look for one hike from each G4 central bank before year-end, markets appropriately price in the risk of earlier and more action. This risk bias is linked to the rising possibility that the Fed starts earlier than our forecast for December. The anticipated 2H26 moderation in US inflation and consumer spending should stay the Fed’s hand until a clearer tightening in labor markets is established. While data surprises may prompt earlier action (we forecast a 4.3% July u-rate next week), this week’s FOMC meeting highlights another reason why the committee may act earlier.

Bruce Kasman · Joseph Lupton · Nora Szentivanyi下载原始 PDF

Expecting a December Fed hike; Sept. hold requires CPI moderation
Impressive balance in 1H26 global GDP as Europe, non-tech activity lift
Japan intervenes while the BoJ keeps door open to September hike
Next week: Global PMI up; US jobs (75k); Banxico holds, COPOM cuts

The three July dissents (we expected two) show a rising number of hawkish FOMC members not likely to be dissuaded by a new chair’s advocacy of patience. Nonetheless, the majority of FOMC voters appear data dependent as they look towards September. Two upcoming employment and CPI reports are central data points, but there is also dependence on market signals around financial conditions and Fed credibility. Ironically, a new Fed chair wanting to divorce Fed meeting communication from market pricing dominated the market response to this week’s meeting. Warsh’s reiteration of his commitment to price stability was undermined by his reticence to consider rate hikes and his failure to endorse the Fed’s well-defined PCE target. He also hinted that market movements could substitute for Fed action, encouraging a bear steepening towards higher real yields. Absent a reversal in these signals, there will be added pressure on the data-dependent majority to counter this move with action in September.

The cycle curving the shocks

With the latest quarter’s GDP reports rolling in, the global economy’s resilience in the face of the latest shock is confirmed. Smoothing out the quarterly China volatility, global GDP rose at a solid 2.5%ar in 1H26 (Figure 1). Although modestly lower than we had anticipated before the closing of the Strait of Hormuz, the sustained resilience across sectors and countries was impressive.

Figure 1. J.P. Morgan global GDP — Figure 1J.P. Morgan global GDP

Figure 2. Non-tech capex and manufacturing — Figure 2Non-tech capex and manufacturing

The global consumer (ex. China) continued to grow at a faster than 2%ar, despite a 4.9%ar surge in consumer prices last quarter. Fears that the energy shock would produce weakness in areas most exposed to the shock , Western Europe and non-tech Asia , proved unfounded. Although the US grew a trend-like 1.8%ar in 1H26, its role as a global demand engine is far greater than this outcome suggests. US import volumes grew at a double-digit rate in each of the past two quarters, spurred on by 15%ar average gains in equipment spending. The indirect global benefits of supportive US financial conditions and a patient Fed are also contributing to positive outcomes elsewhere.

The energy sector disruption has not ended, and geopolitics could still weigh on global growth. Notwithstanding these concerns, our baseline forecast is for a cyclical lift in business spending and hiring to combine with China’s policy stimulus, raising global GDP growth to an above-potential 2.7%ar over the coming quarters. This week’s Politburo meeting guidance supports our view for a “more proactive fiscal policy” in China, which emphasizes faster policy implementation and selective new initiatives.

Tech equipment spending has been an important growth driver, and incoming reports suggest that this demand impulse remains strong. However, recent reports support the view that a broader pickup in spending is taking hold. Asia’s non-tech industrial activity has picked up smartly alongside a rebound in US fixed investment spending (Figure 2). The surge in 1H26 US software spending points to the implementation of tech spending gains across a broader set of companies.

We had anticipated a positive turn in the inventory cycle as part of the rebound in non-tech activity, but available indicators in the US and most of Asia suggest continued weak stockbuilding despite strong final demand. This could be a sign of lingering caution, but we are inclined to link this development to last quarter’s strong final demand that surprised producers and generated tech sector capacity constraints.

The underpinnings for sustaining this cyclical lift in business spending are a normalization in business sentiment from last year’s trade war shock, alongside a profit rebound and supportive credit conditions. Incoming GDP reports point to sustained strong profit gains. Indeed, US 1H26 nominal GDP rose 6.8%ar while labor compensation rose 3.9%ar. Next week’s global PMIs are expected to provide further reinforcement that global growth momentum is building. We look for a rise in the all-industry output index, consistent with improving expectations and solid industrial activity gains.

The search for better income balance

The message from consumers remains upbeat with the US reporting strong 0.4% m/m May and June real gains. Global goods spending outside China is tracking a 4%ar gain in the three months through May. The surprising strength into mid-year contributes to an upside-bias to current quarter growth forecasts that anticipate a material cooling in consumer spending. We nonetheless believe that positive consumption outcomes require better news on real income. A better balance between household purchasing power and profit growth is expected to come from a moderation in global inflation and a pickup in job growth.

Both of these dynamics appear to be underway (Figure 3). Global CPI inflation moderated sharply in June, helped by falling energy prices and a slower core gain. Following a 4.9%ar gain last quarter, we anticipate global CPI (ex. China and Türkiye) to rise 2.4%ar this quarter, even with crude oil prices expected at $86/bbl. The pickup in job growth is anticipated to be more modest , to a 0.7%ar globally and a roughly 100k pace in the US. But its realization may be more significant as it boosts household sentiment and reinforces our view that there are not structural impediments to job growth.

Figure 3. Global inflation and employment — Figure 3Global inflation and employment

The incoming news from the Euro area is encouraging; we estimate a 0.8%ar rise in 2Q26 employment and are tracking a healthy midyear lift in sentiment. There also appear to be positive developments taking place in EM Asia’s tech producers, where a growth boom now appears to be broadening out. This week’s slide in the Conference Board’s July survey suggests US households remain downbeat around labor market outcomes. Against this backdrop, we forecast a 75K rise in July employment next week. While consistent with hours growth rising close to 1%ar over the past three months, it leaves average job gains at a modest 87k.

Domestic demand perking up in North Asia

The AI-driven export boom continues to power EMAX growth, but the key development is that strength is becoming increasingly broad-based. First-half growth has averaged a substantial 5% as domestic consumption and investment join the jump in tech-related exports. Our forecast sees secondhalf growth moderating to a more trend-like basis, but risks are skewed to the upside. North Asia is seeing the strongest gains, and we follow up last week’s upward revision to our 2H GDP forecast for Korea with an upgrade to Taiwan.

Last week’s impressive June Taiwan IP was followed by a remarkable 2Q GDP print which came in at 9.9% ar. While AI-driven exports remains the key catalyst, gross capital formation has emerged as a key growth driver. Private consumption also delivered its strongest annual gain in nearly three years, helped by fiscal support to households. In Korea, IP soared 6.4%m/m last month, joined by large gains in non-tech production. Unlike Taiwan, domestic consumption here remains more uneven. However, notwithstanding the recent equity market sell-off, equity prices are still up more than 150% since March 2025 and should provide a boost to consumers in 2H26.

Varying pressures across DM central banks

Reduced downside growth risks reinforce the case for European central banks to remain focused on inflation. Euro area GDP excluding Ireland accelerated to a 1.3%ar in 2Q26, while Swedish growth averaged 3%ar in 1H26. Consumers appear to have smoothed through the energy shock and business activity has held up better than expected. Against this backdrop, the ECB was on hold this week but seems ready for a second hike in September. July core inflation surprised on the firm side, with core prices rising 0.35%m/m. Alongside stronger growth and ongoing Middle East tensions, the risks are skewed for hikes beyond September.

The BoE also held rates steady this week, though the number of dissents for a hike rose to a 6-3 vote. Governor Bailey and the center of the Committee struck a fairly cautious tone, stressing the lack of convincing evidence that higher energy prices are feeding into broader inflation dynamics. Even so, policymakers acknowledged clear upside inflation risks. We still expect a hike in November, but that will likely require firmer core inflation and further signs that wage settlements for next year are tracking above the BoE’s assumptions.

The BoJ held rates unchanged this week in a meeting preceded by FX intervention, but policy communication shifted in a more hawkish direction. Officials reinforced their willingness to move more quickly, and Governor Ueda maintained a balance by leaving the door open for a September hike while reiterating that underlying inflation has yet to sustainably reach 2%. We continue to expect a hike in October, though the backdrop , of strong industrial activity, firm retail spending, and signs that inflation pressures are broadening , warrants an earlier and larger move than we are forecasting. A September Fed hike could, partly through its impact on the yen, deliver an earlier BoJ hike.

Australia stands apart from the broader tightening narrative. Second-quarter inflation came in materially below the RBA’s forecasts and inflation breadth measures also improved. While policymakers are likely to retain a tightening bias in their guidance for now, this week's data reinforces our view that the hiking cycle has ended and that the next move in rates will ultimately be lower rather than higher.

Fed factor wanes in EM

For EM, the combination of resilient growth, firmer inflation, and a renewed Fed hiking cycle is raising the risk of a more restrictive policy stance. However, the Fed Funds rate itself is only weakly linked to EM policy. What matters more is how Fed policy is transmitted through the USD and US Treasury yields. Even there, EM central banks have become less sensitive than in previous cycles as foreign ownership of local-currency bond markets has declined materially over the past decade. The stronger relationship observed after COVID largely reflected the global nature of the inflation shock, which produced a synchronized policy response across economies. Outside a handful of EMs (e.g.: Korea, Colombia, Singapore), policy rates remain restrictive and significant additional tightening would likely require an abrupt surge in the dollar or Treasury yields. Neither appears to be a material risk at present. Within EM, the sensitivity to US rates is highest in LatAm and emerging Asia (Figure 4).

Figure 4. EM policy rate sensitivity to Fed rate — Figure 4EM policy rate sensitivity to Fed rate

Banxico to hold, while COPOM eases

Domestic fundamentals promote policy divergence in LATAM and we expect Mexico’s Banxico to keep rates unchanged next week following softer inflation and weak activity data, while Brazil’s COPOM is likely to deliver another 25bp cut. However, a Fed hike would pressure regional currencies, increasing the premium on policy credibility. For Banxico, that raises the hurdle for renewed easing. For Brazil, it strengthens the case for preserving optionality as disinflation progresses and the calibration cycle continues.

Table. Global Economic Outlook Summary — TableGlobal Economic Outlook Summary

Table. G -3 economic outlook detail — TableG -3 economic outlook detail

Table. Global Central Bank Watch — TableGlobal Central Bank Watch

Economic Activity Tracking

See here for real-time snapshots and methodology reports.

JPM forecast evolution, GDP and CPI

Our tracking estimate for 2Q global real GDP was broadly steady at 2.3%ar this week. Our forecast for 3Q growth moved sideways at 2.3%ar (Figure 1). Our 3Q26 core CPI forecast ticked up to 2.8%ar (Figure 2).

Figure 2. J.P. Morgan global core CPI forecast — Figure 2J.P. Morgan global core CPI forecast

Table 1. Change in forecast revision indexes %-pt revision to rolling year-ahead outlook over 13 weeks — Table 1Change in forecast revision indexes %-pt revision to rolling year-ahead outlook over 13 weeks

GDP nowcaster and recession probabilities

Our nowcaster estimate of 2Q global GDP growth moved sideways at 2.6%ar, 0.3%-pt above our 2Q forecast (Table 2).

The July DM flash PMIs point to firmer growth momentum, with the output index at its highest since November as services and Western Europe strengthened. Details were also positive: the employment and future output indexes rebounded from recessionary levels, supporting our forecast for a recovery in labor demand. Global export volumes remain around 5% above a year ago, but the early-year surge is cooling, led by EM Asia. Inflation momentum fell in June as energy prices reversed, while global core inflation held near a 2.8% annualized pace. Core services inflation has begun to moderate, but we expect core goods inflation to rise toward 2%.

Table 3. J.P. Morgan global aggregates — Table 3J.P. Morgan global aggregates

Around 60% of the economies we track have seen data surprise stronger over the past 30 days. The nowcaster implies upside risks to 2Q26 growth in about two-thirds of these economies (Figure 3).

Figure 3. GDP nowcaster: risk bias for 2Q26 vs data surprises — Figure 3GDP nowcaster: risk bias for 2Q26 vs data surprises

The US recession probability based on data through July 24 is about 20% (Figure 4).

Figure 4. US one-year recession probabilities since 1995 — Figure 4US one-year recession probabilities since 1995

Capex Nowcaster

Our CapexNow model 2Q26 estimate ticked up to 9.8%ar this week. Observed 1Q capex rose 10.9%ar (Figure 5).

Figure 5. Global capex, actual and nowcaster — Figure 5Global capex, actual and nowcaster

Global consumption tracking

Based on Chase Consumer Card data through July 21, our estimate of the US Census control measure of retail sales is -0.45% in July (Figure 6).

Figure 6. US retail sales control, Census and Chase Card Nowcaster — Figure 6US retail sales control, Census and Chase Card Nowcaster

Global retail sales accelerated to a 2.78%ar in the three months through May. Global auto sales fell 4.3%3m ar in June, but rose 9.9%ar ex. China (Figure 7).

Figure 7. Global consumer spending — Figure 7Global consumer spending

NLP of central bank speak

G -4 central bank communications this week include:

BoJ statement (HDS: -3, trailing 5-speech avg: 0)
BoE statement (HDS: +12, trailing 5-speech avg: -17)
Fed’s Warsh Q&A (HDS: +32, trailing 4-sp avg: +32)
Fed’s Warsh statement (HDS: +49, trailing 3-sp avg: +27)
Fed statement (HDS: +25, trailing 5-speech avg: -7)

Table 4. Hawk-Dove Scores 3m mov avg; -100 (Dovish) to +100 (Hawkish) — Table 4Hawk-Dove Scores 3m mov avg; -100 (Dovish) to +100 (Hawkish)

JPM forecast evolution, policy rates

Figure 8. J.P. Morgan policy rate forecast revision index, DM — Figure 8J.P. Morgan policy rate forecast revision index, DM

Figure 9. J.P. Morgan policy rate forecast revision index, EM — Figure 9J.P. Morgan policy rate forecast revision index, EM

J.P. Morgan forecasts vs. market

Table 5. Policy Rate: J.P. Morgan and Market — Table 5Policy Rate: J.P. Morgan and Market

Real policy rates

Figure 10. Real policy rate and neutral: forecast for 3Q26 — Figure 10Real policy rate and neutral: forecast for 3Q26

See here for NLP methodology and a full set of central bank NLPs. For additional policy rate detail, see our full policy rate QED report.

Table. Selected Recent Research — TableSelected Recent Research

J.P. Morgan Market Watch

Bonds

In the US, short-end yields fell (2-12bp WoW), long-end yields jumped (1-10bp WoW) and the curve twist-steepened (~15bp) after the Fed held rates,an unusually rare post-FOMC pattern. The move was accelerated by Warsh’s remarks that raised doubts about the Fed’s inflation-fighting credibility and commitment to the 2% PCE target. As a result, our call shifts to a December hike (from 3Q27) with risks skewed to further bearish steepening, prompting us to unwind our 5s/30s curve flatteners and initiate 5Yx5Y ZC inflation swap wideners.

In the Euro area, we view the recent steepening Euro curves as unjustified and largely a sympathy move to US Fed-credibility-driven steepening. We see intermediate German yields as cheap and thus remain high-conviction long 10Y Germany outright and long 10Y Bund vs UST as we expect German curve directionality to revert to bull steepening/bear flattening. In the UK, the BoE left rates on hold while turning slightly more hawkish with a tighter 6–3 vote and a wait-andsee stance that retains an implicit tightening bias despite dovish forecasts; we still expect a November hike contingent on firmer underlying services inflation or wage expectations and hold long 10Y gilts. In Japan, the BoJ held rates steady but sounded more hawkish (raising the risk of a September hike if the yen weakens), while GPIF-driven flows, fiscal/tax uncertainty, and mixed super-long JGB demand dynamics leave curve pressure and supply-demand headwinds intact; we hold tactical 5s/20s JGB curve steepeners. (GFIMs, Jul 31st)

In EM rates, we stay cautious amidst ongoing US-Iran escalations and higher energy prices and remain MW overall. We stay OW Hungary, Colombia, Mexico and Paraguay and UW Thailand, Chile and Peru (EM Strategy Update, Jul 24th).

Credit

AI-related credit has had a brutal month, with spreads widening sharply, especially in high-performance compute (HPC) corporate bonds, while securitized data-center paper has been far more resilient. Since early June, HG HPC has widened 38bp to 208bp, and HY HPC has widened 153bp to 418bp, while Data Center AAA CMBS is only ~5–10bp wider amid lighter structured issuance. Data-center project finance has underperformed hyperscalers: HG projects now trade ~99bp wide to related hyperscalers (slightly tighter vs. May), but HY projects trade ~208bp wide (~25bp wider vs. May). The move looks less like market capacity stress and more like investors increasingly dictating pricing and terms as issuance grows (AI Capex Funding, Jul 31st).

LL funds reduced cash from a 4yr high but still hold elevated liquidity and are modestly rotating into riskier credit. Average cash fell 53bp QoQ to 4.35% (still +39bp vs the long-term average) as loan exposure rose to 87.5% and bond exposure slipped to 5.6%; HY funds cut loan allocation to 3.37% (a low since 4Q23). Funds underweighted BBs and NR loans and overweighted Bs and CCCs versus the index, though portfolio yield is 14bp below the index, suggesting similar overall risk. Biggest sector weights are Technology (13.3%), Financials (11.4%), and Healthcare (11.2%), with QoQ shifts led by Healthcare (+71bp) and Services (-62bp). Largest borrowers include TDG and BASSPR, and portfolios tilt toward larger, performing, above-par, private, loan-only issuers (2Q26 Leveraged Loan Fund Analysis, Jul 28th).

Euro IG hyperscaler credit has notably underperformed this month, widening 18bp since 6-Jul (to ~104bp) vs. just 3bp for the broader EUR Senior Non-Financial benchmark, as supply indigestion fears rose after Amazon’s $25bn deal. YTD, hyperscalers have issued about €23.5bn in euro-denominated issuance (plus £5.5bn GBP and CHF5.9bn), but the euro share of total HG syndicated hyperscaler issuance is only 11.4% (GBP 3.1%), below prior expectations as more funding has gone to smaller currencies (CAD, CHF, JPY). Given the lighter-than-expected EUR issuance, the FY26 euro-denominated US hyperscaler supply forecast is cut to €50bn (still ~€27bn likely in Sep–Nov). Hyperscalers are 1.2% of EUR IG today but could reach ~5.7%–7.9% by 2030 (European Credit Weekly, Jul 31st).

Figure 1. Year-to-date returns — Figure 1Year-to-date returns

Currencies

In DM FX, despite a significant selloff post-FOMC this week, Fed action (potentially a December hike), still-supportive carry, and Iran-driven ToT support keep us long USD versus a basket of G10 low yielders (EUR, CHF, SEK, CAD, NZD).

In EM FX, we stay OW overall but remain cautious on renewed Middle East tensions. Regionally, we are OW LatAm (OW: COP, MXN, PYG), MW EMEA (OW: CZK, HUF; UW: RSD) and UW Asia (UW: THB).

Commodities

Iran may aim to influence the Strait of Hormuz not by closing it, but by “regulating” passage through non-discriminatory service fees that it could argue are permitted under international law. Working with Oman, Iran could frame charges as payment for specific services (traffic management, safety, escorts, emergency response, environmental protection) rather than a transit toll, seeking at least tacit UN/ IMO acceptance. UNCLOS generally bars fees for mere passage in natural straits but allows charges for services rendered (Article 26), and similar models exist in the Turkish Straits, Danish Straits, and Russia’s Northern Sea Route. The bigger challenge would be enforcement and collection: illustrative fees could be material (e.g., ~$260k round trip for a very large crude carrier), and dollar settlement would be easiest for shippers,but US sanctions complicate payment, potentially creating leverage for negotiations (Oil Markets Weekly, Jul 30th).

LNG transit via the Strait of Hormuz remains halted, constraining Qatar’s exports and raising winter supply-risk concerns. Bloomberg ship-tracking shows the last LNG vessel exit on July 12 and last confirmed entry on July 10, while deliveries to Kuwait have slowed to just one in the past two weeks. With fewer ballast vessels inside the Strait, QatarEnergy is slowing loadings; we estimate only two cargoes loaded at Ras Laffan last week versus five the week before, alongside 23 LNG vessels still in the Gulf and limited Gulf demand (Kuwait + UAE). Qatar’s implied liquefaction utilization has dropped to ~13% (7-day average) from 22% last week and 30% two weeks ago. Meanwhile, a tighter JKM/TTF spread has flipped netbacks toward Europe, but physical flows are little changed (Global LNG Supply & Shipping Tracker, Jul 27th).

Sugar is the most-exposed agricultural commodity to a potentially very strong El Niño, which could simultaneously disrupt the top export regions,India, Thailand, and CenterSouth Brazil,through different weather channels, tightening global supply. NOAA expects El Niño to strengthen through year-end, with a 97% chance it will last into early spring 2027 and an 81% chance of “very strong” conditions in Q4, potentially the largest event since 1950. El Niño-driven rainfall shifts can alter planting, crop development, and harvest quality. Sugar stands out because these three origins represent ~65% of global exports. India’s risk is largest but often delayed due to irrigation and policy levers; strong events have preceded 17–27% below-trend output, implying 3–5 mt (up to 8–9 mt in very strong cases) at risk vs a 32.5 mt forecast. Thailand’s impact is mixed but very strong events cut ~13–14% (~1.5 mt). Brazil’s risk is smaller (0.6–1.0 mt) but more immediate via wet-season harvest disruption and weaker sugar recovery (Agricultural Commodities Tracker, Jul 15th).

Table. JPM Forecasts — TableJPM Forecasts

The wages of our inflationary sins

Projected cyclical lift to tighten labor markets, slowing
wage disinflation in the DM
Wages play key role in DM inflation via service prices
A post-pandemic steepening in the Phillips curve pro-
vided disinflation with minimal labor market pain…
…but raises risks that the expected labor market re-
acceleration will halt wage disinflation
Productivity should provide some offset, but effect is
more limited in wage-sensitive service sector
DM wage growth stuck at 3% to keep core inflation
above central bank targets

Few are doubting that inflation will remain well above central bank targets for a sixth straight year and likely remain elevated into 2027. Less appreciated is that this has less to do with the headline-grabbing energy shock and pass-through to broader prices and more to do with fundamental supply-de-mand imbalances that are now showing little sign of abating. If the key call we expressed in our year-ahead 2026 outlook is right, the cyclical uplift will continue through 2H26,driven less by the implementation of productivity enhancing AItechnologies and more from a recovery in labor markets. Combined with building supply constraints in both labor and goods production, the demand pickup could keep global inflation stuck near 3%ar through 2H26 and into 2027.

Figure 1. DM unemployment rate and wages — Figure 1DM unemployment rate and wages

This note examines the impulse from labor market pressures on DM core inflation. As the energy shock fades (notwithstanding the recent intensification of the Middle East conflict), the labor market recovery is coming into sharper relief. Against a backdrop of weak labor supply, labor markets are expected to tighten further from already-low levels of unemployment and put a floor under wages (Figure 1).

How much pressure comes from labor markets will depend on two key factors.

• Flattery will get you everywhere. The degree to which wage gains move with tightening labor markets will depend crucially on whether the flatter pre-pandemic (and thus less inflationary) Phillips curve is once again in operation or the steeper post-pandemic (and more inflationary) curve continues (Figure 2). The more recent steeper curve would suggest considerably more wage pressure as unemployment rates drop amid the cyclical upturn in employment expected over the coming year.

Figure 2. Wage Phillips curve, DM — Figure 2Wage Phillips curve, DM

• Productivity offers a helping hand. Productivity can short-circuit the inflationary consequences of rising wage inflation. While DM productivity growth has come off last year’s boil, we expect it to firm again over the coming year and settle around 1%ar. If wage growth runs around 3%ar, unit labor costs would expand at a manageable 2%ar growth pace (Figure 3).

Figure 3. Wages, productivity, and unit labor costs, DM — Figure 3Wages, productivity, and unit labor costs, DM

The case for a flatter Phillips curve is plausible. The rapid wage inflation of recent years may have owed less to supplydemand pressures and more to headline inflation pressures boosting wage demands. With the surge in inflation contained, the old relationship could reassert itself. Moreover, wages could undershoot the Phillips curve, adding to the uncertainty. The case for productivity tempering the wage impulse is less convincing given that inflation pressures are in services, which are more connected to wages than unit labor costs (as illustrated below). Nevertheless, central banks will be sensitive to these issues, and labor market developments in particular, even while recognizing heightened uncertainty.

From labor markets to inflation

After surging in the wake of the pandemic, wage gains in the DM peaked in late -2023 at 5.1%oya and have since fallen rapidly. However, the path of decline is at risk of leveling out above pre-pandemic norms. Our forecast looks for employment growth to double from a weak 0.4%oya in 2025 to 0.8%oya this year and then move up to near 1%oya in 2027 (Figure 4). While we look for some recovery in labor supply as the strength in hiring draws in discouraged workers, it is not expected to keep pace. As a result, the DM unemployment rate is projected to drift lower in the coming year and return to the multi-decade lows seen in 2022.

Figure 4. Labor markets, DM — Figure 4Labor markets, DM

Figure 5. DM core services inflation and wages — Figure 5DM core services inflation and wages

The elevated pace of wage inflation provides a guide to the stickiness of core service inflation. With the unemployment rate moving lower and putting a floor under wage disinflation above 3% annualized, core service inflation should see similar support in the same range. On net, this should keep overall DM core inflation running around a similar pace and above the zone where central banks would feel comfortable.

Ask Phillips, but beware the answer

Central bank policy will rest importantly on the evolution of wages. Wage growth is in part an outcome of the interaction between supply and demand in the labor market. In the pre-pandemic period, the wage-slack dynamic showed wages as relatively inelastic to changes in the unemployment rate. This then changed after the pandemic, as wages jumped faster than many expected.

Figure 6. Wage Phillips curve, US — Figure 6Wage Phillips curve, US

In the US, a 1%-pt rise in the unemployment rate brought down wages by just 0.3%-pt before the pandemic. This socalled wage-Phillips curve relationship was the basis for Chair Powell’s ominous warning in August 2022 that “some pain” in the labor market would be needed to bring inflation back to target. In the event, the wage-Phillips curve had steepened, delivering a wage rate that returned to its pre-pandemic “sluggish wage growth” curve even as the unemployment rate rose only modestly from a multi-decade low (Figure 6).

The steepness of the post-pandemic Phillips curve has been a blessing in delivering a sharp fall in inflation without considerable labor market pain. But this comes with a potential cost as labor markets look to be tightening now, with sizable implications for inflation and thus policy setting. Much hinges on whether the wages move with the pre-pandemic or the post-pandemic Phillips curve. Our forecast looks for the US unemployment rate to drop 0.3%-pt from last quarter to the start of early next year. If the pre-pandemic Phillips curve reengages, then wage inflation will only move up 0.1%-pt. However, if the post-pandemic curve (since 2022) is active, then wage inflation will jump 0.7%-pt,in turn boosting core inflation by 0.4%-pt according to Table 1 below.

Figure 7. Wage Phillips curve, EMU — Figure 7Wage Phillips curve, EMU

The same dynamics can be seen in Western Europe. The Euro area Phillips curve has turned nearly vertical in the post-pandemic period (Figure 7). This has aided in a “painless” disinflation experience. Indeed, despite the unemployment rate being near historic lows, wage inflation has slumped from a peak of 5.6% in 2Q23, to its latest reading of 3.5%. This steepening of the Phillips curve aligns with the ECB’s view that the surge in wage growth in 2022-23 was a response to temporarily high inflation rather than a product of labor market tightness. As with the US, the risk is that the steeper Phillips curve is operative, and any further tightening could also result in a rapid acceleration in wages. Alternatively, wages could just be disconnected from the labor market.

The UK has seen a relatively larger increase in its unemployment rate in recent years, aiding in wage disinflation. The BoE has argued that compositional changes in employment are biasing the average earnings data down by around 0.5%, so the disinflationary impulse may be somewhat more tempered. Nevertheless, the slowing in wage growth has been larger than the pre-pandemic Phillips curve would suggest (Figure 8). As with the US and EMU, this suggests that any cyclical lift that tightens labor markets could transmit more rapidly to wages. This risk is amplified by the evidence from the vacancy-to-unemployed ratio showing tighter labor conditions than implied by the unemployment rate.

Figure 8. Wage Phillips curve, UK — Figure 8Wage Phillips curve, UK

The transmission of labor market tightness to wages is far looser in Japan (Figure 9). Wage inflation has fallen from its 5.2% pace in December 2024, but it remains elevated at 2.8%. Along with the relationship to the unemployment rate being weak, there is no sign of a steepening in the Phillips curve, as with the US and Western Europe. Cost-push pressures have likely boosted wages in recent years, reversing the causality in Table 1 below. As headline inflation settles down, wage inflation should follow. If right, underlying consumer price inflation should continue to drift higher toward the BoJ’s target and allow for gradual rate normalization. 1

Figure 9. Wage Phillips curve, Japan — Figure 9Wage Phillips curve, Japan

Wages at work across G -4 inflation

The primary channel for the pass-through of wage inflation to consumer price inflation is through services. Whereas goods prices are more heavily dependent on external factors,commodity costs, prices of imports, etc.,services are not only traded less but also rely on labor as a larger share of their value-added. (More on this point is discussed in the context of productivity in the next section.) This wage-services inflation relationship broadly holds across the DM, with the correlation between core services and wages being particularly tight in the US and the Euro area (Table 1).

This close relationship has proven an important disinflationary force over the last three years, though the degree to which wages have returned to their pre-pandemic pace varies. The deceleration in wage growth has been greatest in the US. That said, while the Fed has recently characterized labor costs as broadly consistent with 2% inflation, it is still on the higher end, and the risks are skewed upward with a cyclical labor market lift (Figure 10).

Table 1. Wage impact on core inflation Robust regression; Coefficient on wage inflation (t-stats) — Table 1Wage impact on core inflation Robust regression; Coefficient on wage inflation (t-stats)

Figure 10. US core services inflation and wages — Figure 10US core services inflation and wages

Average hourly earnings have also slowed markedly in the UK, from 8.4%ar at its peak in June 2021 to 2.9% in April 2026 (Figure 11). However, the BoE has argued that compositional changes in employment are biasing the average earnings data down. We see the underlying rate running closer to a 3.5-4% annualized pace. To the extent that this is pointing to similar core service inflation, the conditions are not consistent with the BoE’s target.

Figure 11. UK core services inflation and wages — Figure 11UK core services inflation and wages

Euro area measured wage growth is still running well above its pre-pandemic pace, though,like the rest of the DM,it has maintained a steady downward trend from its peak (Figure 12). With headline inflation undershooting the ECB’s 2% target in the decade preceding the pandemic, a return to pre-pandemic wage growth is not inherently desirable. Nonetheless, 3% is still not something the ECB will be comfortable with. To this end, more forward-looking measures are encouraging; both the ECB’s wage tracker and recent corporate surveys suggest wage growth could slip below 3% in the near term.

Figure 12. Euro area core services inflation and wages — Figure 12Euro area core services inflation and wages

The wage-service inflation correlation is the weakest in Japan (Figure 13). Even so, strong wage growth is expected to lift underlying inflation in the second half of the year, keeping the BoJ on a tightening path. The BoJ’s hope is that inflation will gradually drift up to the 2% target. However, the concern is that, with businesses and consumers becoming accustomed to pricing changes, the link between cost pressures and inflation will strengthen and inflation will overshoot absent a steady normalization in policy rates.

Figure 13. Japan core services inflation and wages — Figure 13Japan core services inflation and wages

Productivity less serviceable

Productivity can temper the inflation impulse from a tightening labor market. To the extent that businesses can get more product out of workers, they can also afford to pay more without squeezing margins or raising prices. The recent jump in DM productivity growth to 1.4%ar has helped considerably in keeping unit labor costs in check even amidst high wage inflation (Figure 14). With DM productivity growth expected to run around 1% and wages projected to get stuck around 3%, the increase in unit labor costs should drop back toward 2% over the coming year.

Figure 14. Real GDP and productivity, DM — Figure 14Real GDP and productivity, DM

A 2% run-rate in unit labor costs appears manageable. This would suggest a stable profit margin with 2% consumer price inflation. However, there are a few points of caution. First, whether this then translates into similar gains in pricing will depend in part on the pricing power of firms. Second, to the extent that the wage-to-pricing pass-through is greater for services (where productivity is lower than for goods), the linkage to unit labor costs is muted.

Table 2. DM wage and ULC impact on core inflation Robust regression; Coefficient on %oya inflation (t-stats) — Table 2DM wage and ULC impact on core inflation Robust regression; Coefficient on %oya inflation (t-stats)

The so-called Baumol effect refers to the fact that service industries must raise wages to compete with high-productivity sectors (like manufacturing) for workers, even if service sector workers are less productive.2 To preserve margins, these service sectors need to increase prices in response to rising wages. This is also the reason that some central banks often look at “super-core” inflation (core services ex-housing), as it provides a cleaner read on labor market-driven inflation pressures. While productivity acts as a shock absorber in the goods sector, there is less cushion in services. This is borne out in the data. In contrast to the goods sector, there is a much larger impact of wages on DM service sector inflation than unit labor costs (Table 2). The result is that the inflation pressures coming from labor markets are manifesting in the service sector, where productivity gains (and thus unit labor costs) are playing less of a role.

East Asia: Terms of trade amid tech and oil volatility

East Asia’s terms of trade have been shaped by oppos-
ing forces: a negative energy shock and a positive AI-
driven tech price upcycle.
Korea has seen the largest TOT improvement since
late 2025, while China lags and Taiwan sits in the mid-
dle, reflecting differences in supply chain position and
trade structure.
Tech price support is likely to persist while energy
prices remain volatile, but the translation of nominal
TOT gains into real growth may warrant closer exami-
nation.

Asia has been caught in the cross currents of an energy shock and tech boom. On the one hand, East Asian economies – China, Taiwan and Korea – as major energy importers, saw rising bills after the Middle East conflict; on the other hand, these tech-heavy economies are also witnessing significant export gains led by the AI-driven tech price cycle. Post-conflict price movements in both tech exports and energy imports have been substantial compared to the previous year (Figure 1). This raises the question of which force has had the greater impact on trade income. Terms-of-trade (TOT), defined as the ratio of export prices to import prices, measures the purchasing power of an economy's exports in terms of imports and provides a useful framework for analyzing these offsetting price shocks.

Figure 1. Tech and energy ex/im price moves, Mar-Jun 2026 %oya — Figure 1Tech and energy ex/im price moves, Mar-Jun 2026 %oya

Opposing two price forces in TOT

Recent headline TOT developments suggest that these dualsided price shocks have affected external purchasing power differently across the three economies. Korea has seen the most improvement since late 2025, Taiwan sits in the middle, while China's has weakened the most (Figure 2). Breaking down headline TOT growth by energy and tech sectors helps identify the main drivers behind these divergent outcomes.

Figure 2. Headline terms of trade growth %oya, 3mma — Figure 2Headline terms of trade growth %oya, 3mma

The energy sector has experienced a universal negative TOT shock across all three economies, as seen in 2021-2022 (Figure 3). The movement also tracked Dubai crude oil prices,a useful proxy for energy costs relevant to Asia. This is unsurprising given that all East Asian economies are significant net importers of energy products, with net imports accounting for roughly 10–15% of total customs imports.

Figure 3. Energy contribution to terms of trade growth %pt-contri. to %oya growth, 3mma — Figure 3Energy contribution to terms of trade growth %pt-contri. to %oya growth, 3mma

On the tech side, the picture is more nuanced (Figure 4). China's tech TOT response has been relatively tepid, while Taiwan has recorded a moderate improvement since 4Q25. Korea has been the clear winner, with tech TOT growth accelerating sharply over the same period. The similar timing of TOT gains in Taiwan and Korea suggests a potential common driver,the AI-led tech price upcycle. Importantly, China's muted tech TOT contribution does not imply limited movement in tech prices. As discussed earlier in this note, both China's tech export and import prices have risen significantly. In fact, China’s headline TOT growth has been driven primarily by factors outside of tech and energy.

Divergent tech windfalls in East Asia

The differentiated response to higher tech prices likely reflects two factors: First, economies occupy different positions within the regional tech supply chain. China, as a major manufacturing base for consumer electronics, sits relatively downstream and relies heavily on imported components such as memory and high-end processors from Korea and Taiwan. As a result, it is more exposed to upstream component price increases, while its ability to pass higher costs through to export prices depends on corporate pricing power.

Figure 4. Tech contribution to terms of trade growth — Figure 4Tech contribution to terms of trade growth

In contrast, Taiwan's export mix is more concentrated in high-er-value AI-related hardware, including servers and related equipment, where demand is less price-sensitive and firms possess stronger pricing power. This may help explain why Taiwan, despite also relying heavily on imported Korean memory products, has experienced a more positive tech TOT shock. Korea, meanwhile, benefits directly as a leading producer of memory products and has captured the largest tech windfall, amplified by AI-related demand and supply bottlenecks, compared to the previous semiconductor cycle.

Second, China's much larger and more diversified manufacturing base makes its aggregate export and import price indices less sensitive to AI-related price shocks. AI-related products account for a larger share of Korea's and Taiwan's trade structures, allowing tech price gains to translate more directly into headline TOT improvements. In contrast, China’s broader trade basket dilutes the impact of any single technology cycle. In addition, some imported AI-related components may ultimately serve domestic demand rather than exports, given China’s large domestic tech demand, thereby weakening the pass-through from higher import prices to export prices at the aggregate level.

2H26: Intact tech upcycle, volatile oil

Overall, after navigating multiple price shocks in 1H26, Korea recorded the largest trade-surplus gains among the three economies relative to total customs trade value (Figure 5). Tech accounted for most of the gains in Korea and Taiwan, reflecting positive TOT shocks in the sector, while China's tech and non-tech-non-energy sectors posted gains and losses of broadly similar magnitude. Volume effects also matter. Despite the negative energy TOT shock across all three economies, energy trade deficits narrowed in Korea and China, suggesting lower net import volumes that provided a posi- tive contribution to trade balances as well as a cushioning effect from oil-based product export price increases. Taiwan's tech-surplus gains also appear larger than its tech TOT improvement alone would imply, likely pointing to sizeable volume growth alongside favorable prices. For Korea, tech export volumes in 1H also remained solid compared to last year, combined with the significantly better TOT.

Figure 5. Trade surplus/deficit change breakdown, 1H26 vs. 1H25 % of total customs trade — Figure 5Trade surplus/deficit change breakdown, 1H26 vs. 1H25 % of total customs trade

Looking ahead, the energy-related drag on East Asia’s TOT may linger around given volatile Middle East supply outlook, while AI-driven tech price support is likely to persist as AI capex remains a key pillar of global growth in our mid-year outlook, along with 'higher for longer’ memory upcycle view from our Asia tech analysts (Figure 6). While positive tech-related TOT shocks may continue to support nominal income – including corporate earnings and fiscal revenues, their impact on real production and real net exports is less straightforward. A larger nominal trade surplus does not necessarily translate into stronger real growth if it is driven by pure price gains rather than improvements in product quality, production volumes, or value-added creation. How net exports' contribution to real GDP growth and real trade gains evolve across the region in 2H26 may warrant a closer examination.

Figure 6. Memory price and Korea's export value indices — Figure 6Memory price and Korea's export value indices

South America: Reformdriven capex and growth

South America (ex-Brazil ) may be entering its first
broad-based, reform-led investment cycle in over two
decades...
... as pro-market governments across the region move
to unlock mining, energy and infrastructure
If executed, reform-driven investment could lift poten-
tial growth by 30–70bp over the next decade

South America's largest economies may be approaching their first broad-based, reform-led investment cycle in more than two decades, with the exception of Brazil. A rare alignment of market-oriented governments and pro-investment policy initiatives is taking shape across the region, creating the potential for a meaningful upswing in capital spending. Four developments stand out: Argentina's RIGI framework and the recently expanded "Super RIGI"; Chile's newly approved National Reconstruction and Economic and Social Development Law, the signature reform of President Kast; Peru's Fujimori administration inheriting one of the region's largest mining-investment pipelines; and Colombia's expected reversal, under President-elect De la Espriella, of the Petro-era freeze on fracking and upstream hydrocarbon licensing.

Taken together, announced and prospective investments across these four economies could reach US$250–300 billion over the next decade, concentrated in mining, energy and infrastructure. Argentina and Chile appear to offer the clearest near-term upside. In both countries, the key reforms have largely cleared the political and legislative hurdles that typically delay investment decisions, while sizeable project pipelines are already visible at the project level.

Peru's opportunities are also substantial, though likely to unfold more gradually. Its mining pipeline is among the largest in the region, but fewer than one-third of projects have advanced to detailed engineering or execution, suggesting that much of the headline investment remains years away from deployment.

In Colombia, the incoming administration’s push to revive energy development may ultimately unlock significant investment, but the legal path is far from straightforward. Fracking remains subject to constitutional and judicial constraints that executive action alone is unlikely to overcome. Even so, Colombia's investment story extends well beyond hydrocarbons. After years of weak capital formation, a more businessfriendly policy environment could support a broader recovery in private investment, regardless of how quickly energy projects move forward.

Figure 1. Fixed investment as % of GDP — Figure 1Fixed investment as % of GDP

Regional backdrop: Politics and capital spending align

The political backdrop is reinforcing the investment story. President Kast is already in office in Chile, Keiko Fujimori took office in Peru on July 28 and De la Espriella will be inaugurated in Colombia on August 7. Together, these transitions mark a broader regional shift toward supply-side reforms and market-oriented economic policies.

The key question is no longer whether the political pivot is occurring. It is. The more important issue is how much of the announced investment pipeline ultimately translates into actual fixed investment, over what timeframe, and what that means for potential growth, external balances and mediumterm current-account dynamics.

Argentina: RIGI as a bridge

Argentina's Large Investment Incentive Regime (RIGI), enacted in August 2024, offers approved projects 30 years of fiscal, customs and foreign-exchange stability in exchange for minimum investment commitments. The enrollment window, initially scheduled to close in mid -2026, has been extended, with the possibility of an additional extension at the government's discretion.

According to the latest official data, 20 projects representing roughly US$46 billion have already received approval, while another 23 projects worth approximately US$102 billion remain under review. Together, the pipeline approaches US$150 billion, equivalent to roughly 21% of GDP. The headline number should be interpreted cautiously. Mining and energy megaprojects typically require four to six years to move from approval to completion, underscoring the distinction between announced investments and actual capital spending. Still, the scale is difficult to ignore.

The project mix is unusually diverse. Mining accounts for a large share of the pipeline, with lithium and copper projects concentrated in San Juan, Salta, Catamarca, Jujuy and Mendoza. At the same time, hydrocarbons have become increas- ingly important as RIGI's scope expanded to include upstream oil and gas activities linked to Vaca Muerta.

This combination is what sets Argentina apart. Mining offers export diversification, while hydrocarbons already provide a path toward energy self-sufficiency and eventual LNG exports. Together, they give Argentina both the region’s largest prospective investment pipeline and the broadest sectoral base.

Chile: Turning permits into projects

Chile’s investment challenge is as much about reversing a long decline as it is about creating new incentives. Annual growth in fixed investment averaged 8% between 1996 and 2013, but slowed to just 1.4% between 2014 and 2025. Economic growth followed a similar trajectory, falling from roughly 5% to about 2%. Against that backdrop, President Kast's National Reconstruction and Economic and Social Development Law, commonly known as the RED reform, represents the most ambitious overhaul of Chile's investment framework in years. The legislation amends 36 laws and 15 decrees, combining tax reductions, investment-stability provisions and measures aimed at reducing regulatory uncertainty.

One of its most consequential elements is the creation of compensation mechanisms for projects that receive environmental approval only to see those approvals later overturned by the courts. The provision directly addresses a longstanding source of uncertainty for investors, particularly in mining and infrastructure.

Importantly, of the 63 articles of the mega-reform (38 permanent and 25 transitional), eight of these rules have been challenged by the opposition, related to aquaculture concessions, environmental rating procedures (including the right to compensation for investors) and tax invariability, which would perhaps be the only point of the so-called “heart” that would remain pending. Among the rules that were not the subject of challenges before the Constitutional Court are, for example, the reduction of the corporate tax from 27% to 23%; the tax reintegration; the elimination of the capital gains tax; the reduction of contributions and a limited employment credit. The administration is to bring forward the entry into force of the initiative even though there are allenges from the opposition in the Constitutional Court. However, to do this, first the pending article of needs to be approved by the joint committee and resolve a couple of final observations (vetoes, in legislative jargon) that will be made by the President Kast, issues that would be voted on in a quick movement in the first week of August.

The investment opportunity is large. Chile's official investment registry currently includes projects worth roughly US$314 billion, equivalent to 77% of GDP. Of that total, about US$90 billion is already advancing through environmental review. Recent permitting reforms are also beginning to matter. Law 21.770, enacted in 2025, seeks to reduce approval times by 30% to 70% through streamlined procedures and parallel processing. While separate from the RED reform, the two initiatives work in tandem: one accelerates project approvals, while the other improves the underlying investment framework.

A further boost could come from efforts to expand domestic copper refining and lithium processing capacity. If successful, these initiatives could generate as much as US$100 billion in additional investment over time while strengthening Chile's position in critical-mineral supply chains (Figure 2).

Figure 2. Chile: Projects under environmental review — Figure 2Chile: Projects under environmental review

Peru: A deep pipeline

Keiko Fujimori's narrow election victory paves the way for a decidedly pro-business administration. Unlike Argentina's RIGI or Chile's RED reform, Peru's story is less about creating new incentives than preserving an investment framework that already exists. This matters because the country possesses one of the region's most comprehensive mining project pipelines. The Ministry of Energy and Mines currently lists 66 projects worth a combined US$64 billion, heavily concentrated in copper (Figure 3, see note).

The challenge is execution. Only about 18% of the portfolio has reached the construction or detailed-engineering phase, while another 30% remains at the feasibility stage. In other words, much of the investment still faces a lengthy development process before capital is deployed.

Infrastructure provides another source of support. Since early 2023, ProInversión has awarded more than US$16 billion in concessions, with an additional US$30 billion expected between 2026 and 2028.

For Peru, the key risk is political rather than economic. Mining investments often require five to ten years to move from planning to production. The central question is whether the country's political system can provide the continuity required to turn a substantial project pipeline into realized investment.

Figure 3. Peru mining CAPEX by stage — Figure 3Peru mining CAPEX by stage

Colombia: Beyond the fracking debate

De la Espriella's victory signals a sharp break from Petro’s approach to energy policy. The incoming administration has pledged to reopen upstream licensing and revive discussions around hydraulic fracturing, potentially unlocking up to US$8.5 billion in investment. Yet the legal hurdles are significant. Court rulings and existing restrictions mean executive decrees alone are unlikely to be sufficient. Regulatory reforms, environmental approvals and community consultation processes would still need to be completed before commercial fracking could become a reality.

The bigger story, however, may have less to do with fracking than with the broader investment climate. Fixed investment has averaged just 17% of GDP over the past three years, near-ly five percentage points below its pre-pandemic norm. A combination of deregulation, clearer incentives and fiscal consolidation could help reverse that decline, even if energysector projects take longer to materialize. Even without fracking, the country can reverse the material decline in fixed investment, which tanked by about 6%-pt of GDP in the last six years (see note).

Quantifying the opportunity

Using a standard growth-accounting framework, the reforms discussed above operate primarily through one channel: high-er capital accumulation. To estimate their impact, we apply sector-specific capital-output relationships and adjust headline project pipelines for execution risk.

Table 1. Capex pipeline in Argentina, Chile and Peru — Table 1Capex pipeline in Argentina, Chile and Peru

These are necessarily illustrative calculations rather than forecasts. Nevertheless, they suggest that sustained reform implementation could produce a meaningful lift in potential growth over the next decade. On our estimates, Argentina stands to gain the most, with potential growth rising by roughly 40–70bp over the next five years. Chile and Peru could see gains of 30–50bp, while Colombia's upside is closer to 30–40bp.

Table 2. Translating capex impulse into potential GDP growth uplift — Table 2Translating capex impulse into potential GDP growth uplift

For the first time in many years, South America’s largest economies, with the exception of Brazil, are moving beyond discussion of investment-led growth. They are beginning to assemble the policy frameworks, political alignment and project pipelines needed to make it plausible. Whether the opportunity delivers on its promise will depend less on announcements than on execution. The reforms are largely in place; the challenge now is turning blueprints into shovels in the ground.

Türkiye: Dissecting FX supply-demand dynamics

Although external financing needs are approaching
levels comparable to prior FX stress periods, we view
the FX regime as sustainable in the near term
Gross reserves, at $163bn (including gold), are a
strong buffer against an external financing gap of
$21bn over the next 12 months
In addition, we expect dollarization by residents to
remain contained for now

The cornerstone of the disinflation program launched in 2H23 has been the real appreciation of the Turkish lira. As the current account deficit (CAD) subsequently widened, market participants have, over the course of this year, increasingly questioned the sustainability of the prevailing FX regime. Türkiye’s basic balance – which we define as current account deficit plus net errors and omissions outflows, less net FDI – widened to -4% of GDP, thereby approaching the levels observed during previous currency stress periods (Figure 1).

Figure 1. Basic Balance (CAD - Net FDI + NEO) % of GDP, 12-month rolling — Figure 1Basic Balance (CAD - Net FDI + NEO) % of GDP, 12-month rolling

This note assesses the adequacy of the CBRT’s FX reserves to bridge the shortfall between FX supply and demand and, in the event that pressure were to emerge, seeks to identify the timing and the channel through which such pressure would most likely materialize. We examine this across five channels: the current account deficit, financial flows, external debt repayments, dollarization by Turkish residents, and outflows attributable to net errors and omissions. We conclude that the prevailing FX regime is sustainable for a period of at least 12 months, with an estimated financing gap of $21bn remaining comfortably covered by the CBRT’s FX reserves.

No FX pressure until November

In Türkiye, the monthly current account deficit (CAD) is not evenly distributed across the year. Throughout the summer and until November, we expect the CAD to remain broadly contained. Tourism receipts are at their seasonal peak and energy demand at its seasonal trough, while decelerating domestic demand and cooling credit growth should limit the risk that real appreciation triggers a surge in imports (Figure 2). Moreover, Türkiye’s external debt repayments (aggregate of Treasury, financials, and corporates) are also light over the period from July to October 2026 (Figure 3). As a result, we see no pressure on FX reserves from BoP dynamics until November.

Figure 2. Monthly current account deficit forecast — Figure 2Monthly current account deficit forecast

Figure 3. Monthly External Debt Repayment Projections (Principal+Interest) — Figure 3Monthly External Debt Repayment Projections (Principal+Interest)

We expect this outlook to change going into winter, starting in November. This is attributable in part to seasonal factors, as heating demand elevates energy bills during the winter period. The remaining drivers are specific to the current year. A substantial share of Türkiye's natural gas is procured under contracts indexed to Brent crude oil prices; critically, these contracts do not reference prevailing spot prices but rather the average of prices observed over the preceding six to nine months. Consequently, the elevated Brent crude oil prices recorded in 2Q26 will not be reflected in the energy import bill immediately. Because of this six- to nine-month lag, their impact is expected to materialize only toward the end of 2026. Winter seasonality will subsequently sustain this pressure into the first quarter, rendering the November 2026 to March 2027 period the most demanding stretch of the window – and the interval most likely to weigh on FX reserves.

In 2027, we expect energy prices to moderate and domestic demand to re-accelerate. Our commodities team anticipates that Brent crude oil prices will average $63 per barrel, thereby reducing the energy bill to $45bn from $53bn. Concurrent- ly, we expect GDP growth to accelerate to 4.6% in 2027 from 3.0% in 2026, with the stronger domestic demand pulling in more imports, doubling the core goods deficit to $16bn from $8bn. Two further pressures should build alongside stronger import demand: with gold expected to average $4,775 per ounce, the value of gold imports rises, while a heavier interest burden widens the primary income deficit. The net result is a wider current account deficit of $53bn in 2027 against $48bn in 2026, with its source shifting from energy to domestic demand (Figure 2).

Mind the $21bn gap

On the basis of our current account forecasts set out above, we estimate Türkiye’s total external financing needs at $296bn over the next 12 months from June 2026 to May 2027. This figure combines the $54bn current account deficit with $242bn of external debt due within one year, irrespective of original maturity. We estimate an external financing gap of $21bn over the next 12 months (Table 1). When measured against the CBRT's $163bn in gross FX reserves, the external financing gap is comfortably covered, leaving substantial capacity to absorb the two remaining sources of drain: Turkish residents’ FX demand and net errors and omissions outflows.

Table 1. External financing gap estimation for the next 12 months (bn USD) — Table 1External financing gap estimation for the next 12 months (bn USD)

Net errors and omissions (NEO)

A further drag on the CBRT's FX reserves in recent years has been net errors and omissions (NEO), the balance of payments line that captures unrecorded FX outflows. In Türkiye, NEO has been persistently negative over the past couple of years, signaling unrecorded FX outflows, and reached -$29.9 bn on a 12-month rolling basis as of May 2026 (Figure 4). While the individual components of NEO are inherently difficult to disentangle, three main factors appear to have driven the latest widening in Türkiye's case. The first is profit-taking by foreign investors on long lira carry-trade positions. Because much of this is conducted through FX swaps rather than recorded asset purchases, the outflows from realized gains are not captured in the financial account and surface in NEO. We expect these outflows to persist, though the CBRT's planned BoP revision will record financial derivatives as a separate financial-account line, shifting where they are captured so these flows would be captured there as recorded out- flows rather than surfacing in NEO. The second and third factors – unrecorded gold imports and dollarization by residents outside the banking system – should remain subdued. The attractive lira returns are expected to keep dollarization limited, and weaker dollarization in turn means less gold demand, particularly with no elections expected in the next 12 months.

Figure 4. Net errors and omissions $bn, 12m rolling — Figure 4Net errors and omissions $bn, 12m rolling

CBRT’s FX reserves are sufficient

We conclude that the CBRT has sufficient capacity to sustain the current FX regime over the next 12 months. The CBRT’s $62bn in FX reserves covers the $21bn external financing gap. The CBRT also holds a further $100bn in gold reserves, which can be sold if conditions deteriorate, as demonstrated during the Middle East conflict. Pressure on FX reserves should remain limited through the summer and until November, but then build thereafter as the energy bill rises over the November 2026 to March 2027 window – a period the CBRT's FX reserves are well positioned to absorb.

One risk to this framework is an increase in dollarization among Turkish residents ahead of the next elections. When residents dollarize, however, they largely convert their lira deposits into FX deposits held at local banks, and the CBRT can draw on this FX through swaps with those banks to defend the lira. This extends both the capacity of the CBRT and the duration over which it can defend the current managed FX regime. By way of illustration, during the 2023 presidential and 2024 municipal elections, gross reserves remained above $100bn, supported by the FX swaps with local banks, while net reserves excluding FX swaps declined to as low as -$65bn. Despite this negative net FX position, the CBRT was able to defend the lira. The CBRT`s net reserves excluding FX swaps have since turned positive, standing at $38bn. Accordingly, as long as the FX deposits remain within the banking system, the current managed FX regime can be sustained even under such circumstances.

The authors wish to thank Burak Gok of the EMEA EM Economics team at J.P. Morgan for his valuable contribution to this report.

United States

Chair Warsh’s ineffective defense of Fed credibility
may compel the FOMC to hike by year-end
We see a move in December, but hot inflation readings
could result in a hike as early as September
Conversely, softer numbers combined with June’s low
inflation reading could delay any action
We expect payrolls to rise 75k next week, with a tem-
porary rise in the unemployment rate to 4.3%

The main event of this week was the the FOMC meeting. The Committee left rates on hold, as expected, with three hawkish dissents. However, the big news came in the post-meeting press conference. Warsh’s performance surprised and, based on the market reaction, disappointed many observers,ourselves included. Fed chairs have been hosting these press conferences since 2012, and this week’s was, in our opinion, the most troubling yet by a long shot. Warsh failed to tie his tough talk on inflation to any conditional plan of action. And under the banner of not providing forward guidance, the chair didn’t give any color on the reasoning for the Committee’s decision to remain on hold earlier that day.

Most objectionable, from a central banker's perspective, was Warsh’s casting doubt on the use of PCE as the Fed’s focus for inflation. The chair said the PCE was “the proper standard answer” to the Fed’s preferred inflation measure. But then he added “who knows come after next January what we might say about strategy. I suspect the task forces might have something to add.” This combined two doubts the market has had about Warsh. First, that he might try to “move the goalposts” on inflation, by redefining the target measure. Second, that the task forces are there to rubber stamp the chair’s policy preferences. In either case, the market didn’t like what it heard, with the curve sharply steepening and breakeven inflation compensation rising as the chair spoke.

In our view, the damage to the Fed’s credibility will add some urgency for the rest of the Committee to act to defend the institution’s inflation mandate. Consequently, we have pulled forward our expectation for the next rate hike from 2H27 to December this year, with policy rates on hold at 3.75-4.0% thereafter. There is clearly a risk that the FOMC hikes at the next meeting in September. Between now and then we will receive two jobs reports and two CPI reports. The jobs reports recently have been uneventful, and for next week’s July report we expect a similarly moderate 75k with a likely temporary rebound in the unemployment rate from 4.2% to 4.3%. However, if the CPI heats up the following week, the discussion around the September FOMC would intensify.

The FOMC and its discontents

As noted, this week’s FOMC decision to keep policy unchanged was accompanied by three dissents, which is an uncommonly large number. All three are regional bank presidents (Hammack, Kashkari, and Logan), and all three wanted to raise rates by 25bp this week. Each published a statement explaining their reasons for dissenting, and each contained a number of shared themes: a growing impatience with fiveplus years of above-target inflation and worries that it could become embedded; an assessment that even after accounting for recent shocks, inflation would remain elevated; a belief that the economy is at full employment, so inflation is the more pressing concern; and a worry that avoiding small moves now would necessitate more aggressive moves later. All three dissented earlier in the year over language and have skewed hawkish, so they offer a limited signal as to where the center of the Committee currently sits.

Hammack and Logan also said that they believe current policy is not sufficiently restrictive. Since late last year, our call for the Fed’s next move to be a hike was conditioned in large part on the idea that the rest of the FOMC would,eventually,reach the same conclusion. As noted above, recent events put new urgency into that decision. That said, our forecast threads the case between hiking and holding, and Chair Warsh may appeal to his colleagues to be patient until the task forces deliver some early findings. So we see two-sided risks to the Fed call: that the Committee may need to hike multiple times, and that the data flow will manage to cool just enough to keep policy where it is. We see a greater risk for the former than the latter, and balance those risks with a single hike at year-end. Incoming data should do a lot to clarify the upcoming decisions facing the FOMC.

Slightly softer labor market persists

While inflation has become the top concern for some Fed officials and market participants, the performance of the labor market will also matter for the FOMC. We don’t expect the incoming data to raise concerns about an overheating labor market. Payroll growth slowed in June after a torrid start to the year, and we think the three-month run rate will moderate a bit further in July. The run rate for the weekly ADP employment figures has steadily, if gradually, decelerated over the past two months (Figure 1). One thing we aren’t worried about is a fall-off in employment following the World Cup, whose effects we think are too small to be noticed.

While the jobless claims data continue to signal a robust labor market, other indicators, such as the Conference Board’s labor differential, hint at some upward drift in the unemployment rate. We look for a reversal of the sizable decline in prime-age participation last month, and there is a risk of a repeat of last July’s spike in unemployed new entrants. We thus look for unemployment to edge up to 4.3% for July, before it eventually resumes a very gradual decline. Average hourly earnings should rise 0.2% (3.5oya). This week the 2Q ECI report confirmed that compensation growth is stabilizing at a level higher than before the pandemic,but so too may be productivity growth, a fact some Fed officials have noted favorably in recent months. Productivity growth does appear to have dipped below a 1.0% pace over the last three quarters, but this data is noisy and may yet re-accelerate.

Figure 1. ADP estimate of private payroll growth — Figure 1ADP estimate of private payroll growth

Downside surprise to PCE inflation

June’s flat reading on core CPI was followed up by a moderate 0.13%m/m rise in core PCE, which was not as bad as the 0.22% we had predicted. While easing the immediate pressure on the Fed, this is just one number, and it would take a continued run of softer inflation readings to allow the FOMC to remain patient and refrain from hiking. However, potential new pressures await in goods prices as well as a return to more trend-like readings in ex-housing services inflation. Looking past the single monthly print, core inflation still rose 3.3% over the last 12 months and 2.9% saar over the last three, so by those metrics it remains well above target. The question, of course, is the trend.

Figure 2. Core PCE inflation — Figure 2Core PCE inflation

Recall that at the June FOMC the SEP median had expected core PCE to rise 3.3%Q4/Q4 this year. Inflation could yet come in below that pace, and our reading of the dots was that a majority of voters would be willing to keep rates on hold were that to play out. However, the three dissents at the most recent FOMC meeting, Waller’s hawkish speech just ahead of the June CPI, and the points we raise above about FOMC credibility suggest that the threshold to hold rates steady has likely dropped. Indeed, given that core PCE is already at 3.3%oya, a 3.3% Q4/Q4 reading would imply no progress toward target over the second half of this year.

GDP: final sales run ahead of output

Real GDP growth rose 1.5%q/q saar in 2Q, roughly similar to 1Q once one strips out that quarter’s mechanical lift from the government re-opening. The contours of 2Q growth included a very robust 3.9% gain in private domestic final sales, driven by consumption, equipment investment, and IPP, with a large 1.7%-pt drag from trade and inventories as stored or imported products helped meet that final demand.

One of the more perplexing developments here is inventories, which have fallen for five straight quarters, with much of that coming in the manufacturing sector (Figure 3). This type of behavior is more typical of a recession, and yet goods-producing industries have been performing relatively well, at least in comparison to the standards of recent years. Normally, this type of inventory drawdown would lead us to raise our growth forecast for coming quarters, under the expectation that firms will need to restock, with at least some of that coming from domestic production. We have held off for now while we evaluate why the inventory decline has occurred. The inventory decline has overlapped closely with the trade war, and there is some possibility that an eventual re-stocking might come disproportionately via higher imports.

Figure 3. Real inventories — Figure 3Real inventories

We expect the ISM manufacturing survey in July to rise from 53.3 to 54.0, while we likewise expect the final PMI to hold at 54.0, hardly different from the flash reading. The ISM series has held on to the large spike from the beginning of the year, reaching a cycle-high level of 54.0 in May before inching down to 53.3 in June. The average ISM-weighted composite from the already released July Empire, Philadelphia, Richmond, Dallas, and Kansas City Fed surveys rose to 55.5 from 53.9 in June.

The PMI has noted that some recent strength came from front-loading in response to the Middle East conflict, which could begin to wane. This was mirrored by a drop in the inventory metric in the flash PMI, although supplier delivery times continue to lengthen to the longest in the current cycle. Both input and output price measures in the recent PMI report continued to moderate from recent highs.

While both new orders and production remain at solid levels in the ISM survey, production has been steadily trending down since the January spike. The employment index has been improving since the turn of the year, reaching an 18month high of 49.7 in June . The inventories index in the ISM report has also been elevated in recent months, moving close to levels seen during the Liberation Day period; and a drop in the metric in the coming report could reflect some pullback in anticipatory demand.

We expect construction spending to rise 0.1%m/m in June, the same as May. Private construction remains a drag on total spending, while the public component has been strong in recent months. Housing starts in June recovered from a large drop in the prior month, led by a jump in the volatile multifamily segment, while single-family starts were still down 0.2% m/m, which, coupled with falling prices, is likely to weigh on new residential spending. Looking ahead, elevated mortgage rates and elevated inventories are unlikely to stimulate new construction as builders remain reliant on incentives and price cuts to move existing supply.

Real nonresidential construction has been declining consecutively for over 10 quarters, with 2Q marking a 5.0%q/q drop. We recently noted that the unwind of investments into semiconductor manufacturing facilities, which had earlier been boosted by the CHIPS Act, has been weighing on nonresidential spending. While data center construction remains on the rise, supported by surging AI demand, spending outside of manufacturing remains muted. For June, the steady growth in heavy and civil engineering construction could point to a rise in highway and street construction for the public segment.

Our auto equities team expects light vehicle sales in July to rise to 16.7mn from 16.6mn in June. Much of this reflects resilient demand from consumers despite the uptick in gasoline prices and wider price pressures. The team notes a boost from sales during the July 4th week, along with another surge through the end of the month.

We expect the trade deficit to narrow to $72.0bn in June from $77.6bn the prior month. Much of this reflects the already released advance goods trade report, which saw goods exports fall by 1.8% m/m and imports drop by 2.6% m/m. For services, we expect a strong 1.4% uptick in exports following the solid 0.7% rise in May, boosted by the World Cup from a combination of international travel and tourism spending, along with IP exports from foreign broadcasting rights.

We look for the job openings rate in the JOLTS to fall from 4.6% in May to 4.5% in June. The Indeed measure of daily job openings ticked up slightly between the end of May and June, but over the course of the year it and other private measures had not mirrored a recent uptick in JOLTS, so we expect JOLTS to move back down. The rise in JOLTS job openings was also unusually concentrated in professional and business surveys. Other labor indicators were mixed: the unemployment rate fell in June, but the Conference Board labor market differential continued weakening in both June and July.

We expect the July ISM services composite to inch higher to 54.5 from 54.0 in June. We also look for the final PMI services business activity to remain stable at 53.5 from 53.6 in the preliminary print. The wedge between business activity measures from the two surveys had widened since the end of last year, with the ISM metric signaling a more optimistic message. The new business measure in the ISM survey rose to 60.6 in March, the highest level in over three years, before dropping to 55.1 in June, while the PMI counterpart had been recovering from a recent trough of 49.5 in April to 53.7 in the early June print. However, the PMI press release continues to cite a boost to activity from the World Cup that could fade next month.

While the PMI price indexes had a muted reaction to the beginning of the Middle East conflict earlier this year, the flash July report saw a jump in inflationary pressures. The input price index rose to 62.2 from 59.9 in June, and the output price metric ticked up to 58.9 from 57.8. Meanwhile, the ISM prices index, which saw a 7.7point spike in March, fell in the latest June report to 67.7 from 71.3 in May, although some of the recent escalation in the Middle East conflict could impede this recovery.

We expect nonfarm business productivity to have risen 0.5%q/q saar in 2Q, with unit labor costs increasing 2.1%q/q saar. While quarterly readings of productivity are generally volatile, the resulting 1.9%oya rise would mark the lowest in three years, with particularly soft readings over the course of 1H26. We expect hours worked to grow 1.3%q/q saar, recovering after a sharp 0.2%q/q decline in 4Q25, when hours were dragged down by the self-employed.

We expect initial jobless claims to rise to 210k from 197k in the prior week. The series has been running at unusually low levels, touching the lowest level since 1969 in the payroll survey week of June. We expect claims to inch higher in the coming week, reflecting residual seasonality. Continuing claims continue to move lower and the four-week average is down 7.6%oya, suggesting further eventual declines in the unemployment rate.

We expect that nonfarm payrolls rose 75k overall (and 75k private) in July and that the unemployment rate increased from 4.2% to 4.3%. We also expect that the workweek held at 34.3 and that average hourly earnings rose a high-side 0.2%.

Establishment Survey

Employment growth has averaged around 100k in both the past three and six months, representing an acceleration in comparison to last year’s sluggish 10k average pace. Our would thus represent a slight moderation in the recent pace, though not anything that should prove too worrying.

The most prominent leading indicator that has slowed between June and July has been the weekly ADP data, whose four-week cumulative gain fell from 97k in the June payroll survey week , the week including the 12th , to 60k in the week ending July 11. This has a sizable error band in predicting month-to-month changes in the BLS figure, but we do take signal from this about recent trends in the labor market.

Some other metrics have been more upbeat, especially jobless claims. Initial jobless claims in the July payroll survey week fell to 188k, the lowest since 1969, and well below the level of prior years around the same time. There are large seasonal moves during this time of year and thus the low reading could reflect seasonal adjustment challenges, but it was notable that claims remained unusually low in the following week as well.

Among the business surveys, the regional Fed surveys have in both of the last two months been stronger than the prior couple years. The PMI all-industry employment index has also risen, though for that survey the gain is less impressive, since it only finishes reversing a large May drop.

Daily employment data from Homebase has meanwhile been a little bit soft, and probably suggests a bit slower growth than the recent trend.

Another drag on July employment could be World Cup employment, assuming there started to be some moderate reduction in employment as the number of matches dwindled. However, we doubt the effect is large enough to matter. Nothing has stood out to us in the industry breakdown of employ- ment so far, and likewise we don’t see excess employment growth in World Cup cities being reported ion the BLS’s MSA employment data.

Government employment has recently turned positive again, though we expect no gain in July. Local government non-education employment jumped 28k in May and then eked out a small gain in June, with more detailed segment data through May showing the rise was concentrated in general administration. This might have been linked to poll workers, and if so could pull back in July since there were no primary dates in the first half of the month.

Seasonal considerations

We believe the early timing of Memorial Day likely boosted leisure and hospitality employment in May (+40k) and depressed it in June (-61k). That left the actual level of leisure and hospitality employment the lowest since February despite a possible lift from the World Cup. Employment now looks low relative to trend in the sector and growth should firm up in July. Leisure and hospitality rose an average of 10k per month in the last year, and fell 2k per month in the last six months.

There is also a general concern that summer employment growth may have weakened over the last few years due to changing seasonal employment patterns. We've found this hard to validate since the weakness hasn’t always shown up in the same sector or month. As we discuss in the Household Survey section below, there has been a pattern of recent college graduates seemingly taking longer to find employment during the summer months, which could create a pattern of summer weakness that doesn't always appear in the same industry or month. In any case, July was a weak month in 2024 but not in 2025, with August being the month that has been weak in both years.

Revisions

The revision pattern has recently turned more balanced, so the consistent pattern of downward revisions in earlier years may no longer hold. That being said, May and June have tended to see downward revisions at the time of the July report, whereas July itself doesn’t have a clear directional bias to revisions.

Hours and earnings

The workweek has held at 34.3 for three straight months, the first time it has maintained that level for more than two consecutive months over the last couple years. The strengthening job market could be starting to lead to slightly better hours figures, and we expect the workweek to hold again at 34.3.

Average hourly earnings rose 0.35%m/m in June and have averaged 0.26% in the last three months. The number of weekdays in the month can affect reported earnings growth, with more workdays leading to lower hourly earnings. This July had 23 weekdays, the most possible, which in the past has meant earnings growth centered around the trailing average. Given the strong June reading we look for a small pullback in July, and forecast a high-side 0.2% reading.

Household survey

The unemployment rate was 4.2% (4.189%) in June, the low-est in a year. We expect the rounded unemployment rate to rebound to 4.3% in July, even as the ongoing moderation in continuing claims suggests that unemployment will return to a downward trend in coming months.

Two technical factors could be in play for July. First, the participation rate plunged 0.3%-pt in June, with much of that concentrated in an unusually large decline in the 25-34-year-old segment. We suspect this is mostly noise and participation will rebound. That doesn’t necessarily imply a mechanical increase in unemployment, but it does probably create upward pressure on the unemployment rate. We look for participation to rebound at least 0.2%-pt, to 61.7%. Offsetting some of the rebound in the 25-34-year-old group will be some residential seasonality in the 16-24-year-old segment which causes participation to fall in the summer.

A second factor to consider is that there was an unusual spike in unemployed new labor force entrants in July of 2025. This did not occur in prior years so we don’t know that this is a reoccuring issue. However, it could highlight challenges seasonally adjusting the flow of summer job seekers, or a genuine increase in joblessness among that group.

A separate point to highlight is that the unemployment rate among young college graduates has shown more of a pronounced summer hump over the last few years, with this year’s pattern tracking a similar path. This will tend to create upward seasonal pressure on the unemployment rate.

Away from these considerations, leading indicators are mixed, in the same way they have been in recent months. As before, the most upbeat signal comes from continuing claims. They are now in line with their level in 2023, a period when the unemployment rate was still near the mid -3% range. As we discussed recently we don’t expect unemployment to immediately return to such a low level because of still high long-term unemployment, but the signal here is that unemployment should fall over time.

On the other hand, the labor market differential and the NFIB survey have both continued to weaken. Both of these have proven to be successful contemporaneous (labor market differential) and leading (NFIB) indicators of unemployment. Nonetheless, continuing claims have better called the drop in unemployment that began late last year.

Review of past week’s data

Total durable goods orders rose 0.3% in June, as the civilian aircraft segment rebounded 3.7%. Ex-transportation orders firmed 0.6%, and core capital goods orders,nondefense ex aircraft,increased 0.9%, on top of an upwardly revised 1.9% May gain. Durable goods shipments rose 0.7% overall, 1.0% ex transportation, and 1.9% for core capital goods, the largest monthly gain since late 2021. Core capital goods shipments expanded 11.1%ar in 2Q, even stronger than 1Q’s 8.4% pace.

In the June advance goods trade report, exports fell 1.8% and imports declined 2.6%. This was the second straight fall in exports though this comes after brisk growth in earlier months. The decline in June was a bit concentrated in industrial supplies, which can be affected by prices and volumes for petroleum and precious metals, and excluding that exports were flat. The drop in imports was widespread, affecting every major category. The capital goods deficit narrowed $2bn on the month, but is still much wider for 2Q overall, driven by surging tech imports.

Consumer confidence inched down in July to 90.8 from 92.2 in June, as the present situation index continued to deteriorate and is now at its lowest in the current cycle. The press release noted that “Comments about food and grocery prices increased,” and while comments on war and energy prices were muted, they could increase in the final survey given the higher gas prices near the end of the month. The expectations index remained steady at 74.7, moving in a stable range since the summer of last year. The most concerning signal from the report was the muted print on the labor market differential, which saw its June value revised from a cycle-low 2.4% to 3.8%, before falling to a new cycle low of 3.1% in July.

2Q real GDP rose 1.5%q/q, saar and the broad story here is that private domestic final sales were strong (3.9%), driven by consumption (3.2%), equipment (15.2%), and IPP (8.8%), though with an offset primarily from trade and inventories. Trade and inventories together took off 1.7%-pt, matching our forecast collectively, though with more of that coming from inventories (-0.7%) and less from trade (-1.0%) than we had predicted. This now marks the fifth straight quarter that inventories have declined, including a $51bn annualized drop in 2Q. On equipment investment, there was a pause in computer investment, which fell 3.5% after a huge runup in earlier quarters. Weaker parts of the report included government (-0.8%) and nonresidential structures (-5.0%). The BEA noted that the drop in federal nondefense spending, which drove the government decline, was a function of sales from the Strategic Petroleum Reserve. For nonresidential structures, the decline in 2Q is part of a recurring pattern, as this segment has been falling for 2.5 years straight.

The June core PCE deflator rose just 0.13%m/m (1.6%m/m ar, the lowest since March 2025). The market-based core PCE deflator was slightly firmer at 0.18%m/m, though both the regular and market-based core rose 2.9% saar in the last three months. All the key categories rose at a similar rate to the overall core PCE in June, with goods increasing 0.09%m/m, services 0.15%, and services ex housing (supercore) 0.12%.

Consumer spending remains strong, though that continues to outpace disposable income, leading to an ongoing fall in the saving rate. Real consumption rose 0.4%m/m, slightly more than the 0.3% rise in real disposable personal income.. The saving rate fell to 2.7%, which is now down 0.9% in just the last six months.

The employment cost index (ECI) showed total hourly compensation expenses rose 0.9% (3.6%ar) between March and June. Except for a brief dip in 2H25 the quarterly rates have held in a narrow range over the last couple of years, and the over-year-ago rate has now held at 3.4% for three straight quarters. The over-year-ago rate of growth for wages edged down again, from 3.3% to 3.2%, which especially for privatesector workers (3.1%oya) puts things back nearly where they were just before COVID. It is thus benefit costs that are holding total compensation above pre-COVID rates. The over-year-ago growth in benefit costs was 3.8%, the most since the start of 2025, and up from a recent low of 3.4% at the end of 2025 The final University of Michigan consumer sentiment print for July inched higher to 55.2 from 54.4 in the preliminary report. Much of this improvement came from the expectations index, which rose by 1.4 points to 55.4 in the final reading, while the current conditions measure inched down by 0.1 points to 54.8, although both indexes remain above their June prints. Inflation expectations held steady from the early report and were slightly lower than in June. One-year inflation expectations are down to 4.2% in July from 4.6% in the prior month, and the five-year-ahead expectations remain steady at 3.3%.

US Focus: When will nonres. construction stabilize?

Nonresidential construction fell 5.0%q/q saar in 2Q, the tenth straight quarterly decline (Figure 1). That nonresidential structures spending is continuously falling is perhaps surprising when data center construction is soaring. Moreover, we expect a continued decline in the near-term, as spending related to the CHIPS Act continues to drop. By next year, though, we should get some tepid growth again, and we look for non-residential investment to rise 2% in 2027.

Figure 1. Real nonresidential structures investment — Figure 1Real nonresidential structures investment

A useful starting point is to look at two parts of the technology buildout together: data centers and computer manufacturing. The AI boom is, of course, driving growing demand for data centers. But the wind down of spending on computer manufacturing projects that were supported by the 2022 CHIPS Act is moving in the opposite direction. The nominal level of these two categories are now about equal, but the absolute dollar decline in computer manufacturing facilities is much greater than the growth in data centers, so their combined change in spending is still negative (Figure 2). Note that data center construction spending excludes the IT equipment inside of them, which is part of equipment investment.

Figure 2. Real data center and computer manufacturing construction — Figure 2Real data center and computer manufacturing construction

If the last year’s rate of decline persists, spending on computer manufacturing facilities will fall back to their pre-CHIPS Act level within the next year, at which point this drag will abate, though it could be more drawn out if a slower decline observed in 2Q persists. It’s also worth pointing out that this decline is not a sign of weakness, but rather the natural counterpoint to the huge surge in construction that began in 2021. And as construction finishes up, output can rotate from putting up buildings to making semiconductors.

Even after computer manufacturing construction stabilizes, this won’t necessarily translate into quick growth in overall nonresidential investment. Nonresidential structures investment excluding manufacturing has recently been close to flat, falling 0.9%oya in 2Q, though that’s a much smaller rate of decline than persisted through early last year (Figure 1). Spending is still falling because while data centers are growing , up 18%oya , they account for less than 6% of total nonresidential construction. Spending on commercial and healthcare buildings excluding data centers fell 4.5%oya, held back by high office vacancy rates and elevated mortgage rates (Figure 3), with both those headwinds likely to persist.

Figure 3. Commercial/healthcare cons. ex data centers, office vacancy — Figure 3Commercial/healthcare cons. ex data centers, office vacancy

Among other major groups, only power/communication and mining are up, though not by much. Power and communication should also be benefitting from data center demand, though real spending only rose 1.2%oya (Figure 4). Note some parts of power investment will appear in equipment.

Figure 4. Real power/communication, mining structures investment — Figure 4Real power/communication, mining structures investment

Euro area

Euro area GDP grew 1.8%q/q saar 2Q26, with
0.5%-pt of this due to distortions from Ireland
Ex-Ireland growth of 1.3%q/q saar shows resilience to
the energy shock and creates upside risk to 2H26
Euro area core inflation looks stickier again after
upside surprise in July
ECB outlook remains hostage to the energy crisis, but
risk remain skewed to another hike after September

This week delivered key growth and inflation data. On growth, the 2Q preliminary GDP release confirmed the region has remained resilient despite heightened geopolitical uncertainty and the energy shock, as it has for the past two-and-ahalf years. GDP rose 1.8% q/q saar, with only 0.5%-pt due to Ireland-related distortions; ex-Ireland growth was 1.3% q/q saar,above potential,despite the sharp purchasing-power hit from higher energy prices. The Euro area PMI dropped sharply at the start of the Iran conflict but rebounded in July to its pre-war level (Figure 1). There is a risk that next week’s final PMI is revised down as tensions increased over the course of the month, and the final survey will incorporate responses from the latter part of July. If not, the survey would imply growth running near a ~1.5% ar pace.

Figure 1. Euro area real GDP — Figure 1Euro area real GDP

On inflation, this week’s flash HICP report for July showed an 0.35%m/m, sa rise in core prices. This left core inflation momentum looking stickier again, at ~2.5%ar (Figure 2). The full details in the final release in mid-August will help to gauge the sources of this: a sticky pre-war trend, energy passthrough, rising tech prices, noise in July or something else. Our view on these sources has moved around this year. We entered the year with a benign view on inflation, with incoming wage data still supportive of this. The actual inflation data were firmer, however, in the Jan-Apr period, suggesting greater pre-war stickiness than we had expected, especially as signs of energy pass-through were modest. Fading drags from tech prices likely contributed as well. The May-June data on core were softer again, but that impression is now being reversed by the July report. That core is still running above target would be consistent with the pricing surveys from the business surveys, which have moderated from their recent peaks but are still at elevated levels.

Figure 2. Euro area core inflation — Figure 2Euro area core inflation

The ECB’s narrative has also changed in recent months, from seeing clear signs of indirect effects from energy to core to seeing a softer trend in June. The July report will likely rebalance that as well and reinforce the case for a September hike, with ongoing tensions in the Middle East and significant upside surprises to its recent growth view also contributing to that. With September feeling highly likely, the bigger question is about what happens thereafter. Our forecast has September as the final hike. Risks are skewed to one additional rate increase thereafter, but this remains somewhat dependent on how the situation in the Middle East evolves. Despite low storage levels, Brent and TTF still tend to fall quickly when hostilities recede.

Core inflation stuck at 2.5%ar

Euro area headline and core inflation ticked up 0.1%-pt to 2.9%oya and 2.5%oya, respectively (Figure 3). This was in line with our expectations on headline and was 0.1%-pt high-er on core, although our forecast was close in the rounding.

Figure 3. Euro area HICP inflation — Figure 3Euro area HICP inflation

Energy inflation rose 1.5%-pt to 10.0%oya as new tensions in the Middle East pushed up Brent (and hence fuel prices) during the month and as some energy support measures expired in some countries. Further pass-through is likely in August unless Brent declines again. Food inflation continues to surprise on the downside, slipping another 0.3%-pt to 1.2%oya, mostly reflecting moderation in unprocessed food inflation. We note, however, that pass-through from energy to food can take quite a while and, therefore, some upward pressures may still show up. In the ECB staff’s forecast, the peak impact from the energy shock on food inflation is not until 1H27. Goods (0.41%m/m, sa) rather than services (0.33%m/m, sa) drove the upside surprise on core. This pushed up the annual rate of core goods inflation by 0.2%-pt to 0.9%oya, while services inflation increased 0.1%-pt to 3.3%oya. The latter was a move of only 4bp, however.

Sentiment rise reinforces PMI signal

The EC survey reinforces the message from the flash PMI that the growth backdrop is improving. The EC sentiment survey was particularly encouraging, with the headline ESI rising 1.5pts to 96.9 and the improvement broad-based across both sectors and countries (Figure 4). The detail fits our interpretation of the recent hard data. Euro area households appear to have smoothed the latest real income shock, helping to sustain spending through 2Q, with goods demand, including autos, proving particularly resilient. The latest survey evidence suggests that any consumption slowdown was temporary. Consumer confidence remains low in absolute terms but improved further in July (+1.7pts m/m), consistent with household spending gaining traction at the start of 3Q. Taken together with the other July releases available so far, the survey data lift our 3Q household consumption nowcast to +1.7% ar, from +1.3% prior to the PMI release and 0.9% in late June. It remains early days and hard data are not yet available, so the signal should be treated cautiously. Still, taken at face value, the latest evidence points to a meaningful upside risk to our current +0.75% ar consumption forecast.

Figure 4. Euro area EC economic sentiment and PMI — Figure 4Euro area EC economic sentiment and PMI

Labour market remains source of resilience

The resilience in household spending continues to be under-pinned by an equally resilient labour market. Despite the size- able energy shock, firms do not appear to have materially pulled back on hiring, with employment growth tracking an acceleration from 0.3%ar in 1Q to ~0.8%ar in 2Q, broadly in line with our forecast. June unemployment data reinforce that message. The Euro area unemployment rate held steady at 6.3%, remaining near its historical low for over a year and a half (Figure 5). While the number of unemployed increased by 71,000 over the month, this was largely driven by Italy, where data is very volatile, making monthly readings noisier.

Figure 5. Euro area unemployment — Figure 5Euro area unemployment

That said, the German labour market remains a pocket of weakness. The latest data showed employment fell by 23,000 in June, bringing the cumulative decline in employment in 2Q to 68,000 (-0.5% q/q saar, Figure 6), versus a 48,000 drop in 1Q. Meanwhile, the number of unemployed rose by 6,000 in July, pushing the national unemployment rate up 0.1%-pt to 6.4%. Forward-looking indicators offer little reassurance: vacancies were unchanged in July, remaining near recent lows, and the flash employment PMI, despite improving, points to continued weakness in hiring at the start of 3Q. Taken together, the data suggest labour market conditions will continue to lag the broader improvement in activity in the near term. The silver lining remains the cyclical lift in productivity: The 0.9% ar 2Q GDP expansion combined with a 0.5%q/q saar contraction in employment implies a +1.4% q/q saar rise in output per worker, leaving labour productivity growing at an average 1.2%ar pace over the past year.

Figure 6. Germany employment — Figure 6Germany employment

Data releases and forecasts

Week of August 3 - 7

German orders have been impacted strongly by lumpy flows of defence and other bulk orders. Such orders are recorded in full in the month they are placed and therefore cause spikes in headline orders. These spikes are noisy but should not be ignored as they cumulate to a large orders backlog for German industry. Ex-bulk orders have also moved higher, however, giving a picture of an underlying pickup. Both export orders and domestic orders have increased. These developments have run ahead of slower improvement signaled by the German manufacturing PMI. For June, we have penciled in a temporary pause in total orders and another increase in ex-bulk orders.

The April/May average for German IP was already 2.1%ar above 1Q26, which included a rebound in construction that was partly due to the weather. The VDA report for June showed car production increasing and both total orders and factory sales in manufacturing have shown an even firmer trend through May than IP. The German manufacturing PMI has also moved back to signaling underlying growth in IP. All of this points to further growth in IP.

French manufacturing output declined in May, with broadbased falls across sectors. Some recovery from this is likely in June, although the PMI was still signaling a weak underlying trend.

Through 2Q, the signal for household spending on goods (retail goods and autos) has been very robust. We expect Euro area retail sales to have increased further by 0.2%m/m in June, likely boosted by fuel sales as fuel prices continued to come off in June. Country-level detail available so far remains limited, but data in Spain shows a 0.3%m/m rise. Our forecast, however, embeds some payback in Germany after a very strong May.

Retail sales rebounded by 1.1% m/m in May, partly driven by a recovery in fuel sales following the government’s tax cuts aimed at cushioning consumers from higher energy prices. Even so, fuel sales remained below pre-shock levels, leaving scope for further improvement as the measures were in place through June. That said, other components of the May report were also robust, and we expect some payback in June.