Where the services jobs went

• The non-US share of the stack keeps growing

  US services employment has roughly doubled over the span shown, but the destinations have grown ~3 to ~6 times. The shape of the stack (the US band staying roughly the same thickness while everything above it expands) is what the cost-arbitrage story looks like in aggregate labour-force data. The US didn't lose; the rest of the world built much faster.

• India and China are not the same story

  China's services workforce growth tracks its urbanisation curve more than its offshore-receiving role. Most of that mass is domestic retail, transport, food, public administration. India's growth has a much larger remote-services / business-process component (visible if you compare India's services sector vs. its overall labour force; the share rises faster than China's). The offshore narrative fits India cleanly; for China it's a smaller slice of a larger story.

• The arbitrage destinations are mid-sized but additive

  Mexico, Poland, the Philippines, Vietnam, and Brazil are each smaller than the US or the two giants, but stacked they form a meaningful slab. The Philippines' services band has more than tripled over the window, largely on BPO buildout. Poland's growth tracks the Eastern-European tech-services hub story. Vietnam and Mexico are slower but steady.

• Real shocks are visible in the stack

  The early 2000s dip after the dotcom bust shows up in the US band. The 2008/09 GFC is a flatter year for several countries. The COVID notch in 2020 is the sharpest collective dip, and the recovery in 2021/22 is steepest in destinations whose services sectors were growing fastest into the shock.

• "Services" is wider than "offshorable services."

  ILOSTAT's services aggregate includes everything from software engineering to street vendors to nurses. Most of an Indian services worker's contribution to this chart is domestic, not work for a US client. The trend is real and directionally correct for the offshoring story; the absolute counts overstate the "jobs that competed with US jobs" slice by a wide margin. A future cut on ISIC J (information & communication) and M (professional services) alone would tighten the framing.

The table is locked to the most recent year-pair any selected country has data for (the US publishes 2025 estimates well before most others). Countries whose latest ILOSTAT release doesn't reach that pair are excluded from this table; they still appear in the main chart. This keeps every row comparable on the same window without burying the latest US number.

• SRC Numbers come from ILOSTAT indicator EMP_TEMP_SEX_ECO_NB_A (employees by sex and economic activity, annual, in thousands). We pull the ECO_SECTOR_SER (services) value as the numerator and ECO_SECTOR_TOTAL as the denominator for the share.
• SCALE Raw counts of services-sector workers, stacked. Y-axis is in people; the top of the stack is the total services workforce across all selected countries combined. The US dominates the absolute mass; the story is in how much faster the non-US bands grow from the early 2000s onward.
• SCOPE "Services" here is the ILOSTAT ECO_SECTOR aggregate (everything that isn't agriculture or industry). That bucket is wider than just the offshorable tech / BPO / professional-services slice; it also includes retail, hospitality, transport, public administration. Trends are directionally honest for the "jobs moved overseas" story but the absolute counts overstate the strictly-arbitrage slice.
• GAPS Several destinations have patchy ILOSTAT coverage in the 1990s. A line that starts late just means survey coverage started late, not that the workforce was zero.