Interactive two-sector model of unbalanced growth — why services keep getting more expensive while goods get cheaper
Baumol's 1967 model divides the economy into a progressive sector (manufacturing, tech) with rapid productivity growth and a stagnant sector (education, healthcare, arts) with slow growth. In the classic case, wages in both sectors rise at rate gP because labor markets are integrated. This page generalizes that setup by letting wage growth w vary separately, so progressive-sector cost follows e^(w-gP)·t and stagnant-sector cost follows e^(w-gS)·t. When w = gP, the classic Baumol result is recovered: progressive-sector cost stays stable while stagnant-sector cost rises as e^(gP-gS)·t.
Performing a Beethoven string quartet requires 4 musicians and ~30 minutes in 1820, 1920, and today. Labor productivity is essentially zero growth. Yet musicians' wages must track the broader economy. You cannot "speed up" the performance without destroying the product. This applies equally to teaching, nursing, therapy, caregiving — services where human time IS the product.
If society maintains a stable service-to-good consumption ratio, labor must flow from progressive to stagnant sectors: Ls/Lp ∝ e^(gP-gS)·t. All developed economies have become service-dominated — not from preference shifts, but from differential productivity growth.
Healthcare is the largest cost-diseased sector. Doctor consultations, nursing care, and therapy require sustained human attention. Even with advanced technology, core clinical encounters remain time-intensive. Research confirms Baumol variables significantly explain healthcare cost escalation across US states and OECD countries.
University tuition has outpaced inflation in nearly every developed country. Core teaching — discussion, mentoring, feedback — has near-zero productivity growth. Online courses reduce distribution costs, but high-quality educational relationships remain expensive. AI will automate grading, but personalized mentorship may command an increasing "Baumol premium."
As populations age, eldercare becomes the fastest-growing cost-disease sector. Personal care, companionship, and daily living support require human presence. There is no Moore's Law for empathy. Eldercare costs are projected to consume an ever-larger share of GDP worldwide.
Cost disease is the logical consequence of differential productivity growth in an integrated labor market — not corruption or waste. Goods get cheaper because manufacturing is so productive; services get relatively more expensive because they are not. The problem is distributional: lower-income households feel the squeeze most.
AI will reduce but likely not eliminate cost disease. It can automate administrative tasks and routine interactions, but clinical judgment, emotional support, creative performance, and trust-building remain resistant to full automation. Basic services may become more accessible through AI, while premium human services become even more expensive.
Governments face inexorable cost pressure in healthcare, education, and social services. More promising approaches: invest in genuine service productivity, use AI for routine tasks while preserving human judgment, expand social insurance, develop mixed delivery models, and focus on equity.