Poor economics in the FT

The FT gets Baumol’s disease wrong: Productivity growth raises the relative cost of labour-intensive services. It also raises the income available to pay for them.

Chrystia Freeland’s recent Financial Times column uses Baumol’s cost disease to explain a familiar complaint: voters feel that they are paying more tax while receiving worse public services. Her description of Baumol’s mechanism is broadly correct. The fiscal conclusion she draws from it is not.

Freeland argues that labour-intensive services such as teaching, healthcare and social care cannot match productivity growth in sectors producing manufactured goods. A string quartet still requires four musicians for the duration of the performance; caring for a child or an elderly person still requires human time. Yet wages in these activities must broadly keep pace with wages elsewhere, or their workers will leave. Their cost therefore rises relative to the cost of goods produced in sectors with faster productivity growth. This is the basic mechanism described in Baumol’s 1967 paper.

So far, so good. Freeland then suggests that one response is to “levy higher taxes on the sectors that can become more productive”. That leap does not follow. Productivity growth elsewhere raises the cost of maintaining public services, but it also raises the national income from which those services are funded. If we want the same quantity of public services and they continue to employ the same share of the workforce, the public-service bill and the tax base rise together. The tax share need not rise at all.

The arithmetic

Take a deliberately simple economy with GDP of £100. The private sector produces £50 of goods and the public sector provides services costing £50. Public spending and tax revenue are therefore both 50 per cent of GDP.

Now suppose private-sector productivity rises by 10 per cent while public-sector productivity is unchanged. The population and workforce are fixed, and people want the same quantity of public services as before. Providing it therefore requires the same number of public-sector workers.

To retain those workers, public-sector wages must rise in line with private-sector wages. Let's keep the price per unit produced by the private sector constant, then wage rates in both sectors must rise by 10 per cent (to match the 10% rise in private sector output quantity). The result is:

	Before	After
Private-sector output, at current prices	£50	£55
Public services, valued at current cost	£50	£55
Nominal GDP	£100	£110
Tax revenue required	£50	£55
Tax revenue as a share of GDP	50%	50%

The same public services now cost £55 rather than £50. But nominal GDP has risen from £100 to £110, so an unchanged tax-to-GDP ratio raises exactly the additional £5 required. Freeland counts the rising public-sector wage bill but neglects the simultaneous growth of the income used to pay it.

Valuing public services at their current cost is not an accounting trick invented for this example: in the absence of a market price, national accountants generally value current-price non-market output using its costs of production.

The distinction between nominal and real GDP also makes clear what Baumol’s disease actually does here. At the original prices, private output has risen from £50 to £55 while real public-service output remains £50. Real GDP is therefore £105: a rise of 5 per cent. Nominal GDP is £110, implying GDP-deflator inflation of about 4.8 per cent. The implicit price of public services has risen relative to the price of private goods. That is Baumol’s cost disease. The share of society’s labour and income devoted to public services has not risen.

If “higher taxes” merely means that the government must collect £55 rather than £50 after nominal income has grown by 10 per cent, the statement is true but empty. Governments normally collect more pounds when the economy’s cash income grows. If it means a higher tax rate or a larger tax share of GDP, it is false - as this example shows.

The article’s language about “makers” and “takers” is also misleading. Public-sector workers are producing education, healthcare and care services; taxation finances their collective purchase. The economic issue is how labour is allocated between different forms of production, not how much the supposedly productive must surrender to the supposedly unproductive.

What could make the tax share rise?

There are several reasons why an economy might genuinely need to devote a growing share of its income to public spending. An ageing population may require more healthcare and social care. People may demand more or better public services as they become richer. New medical treatments may expand what health systems can do, while increasing their total cost. Public-sector productivity might fall rather than merely remain stagnant. Debt interest, pensions or other spending may displace money previously available for frontline services.

The distribution of private-sector gains also matters. If productivity growth appears mainly as lightly taxed or internationally mobile profits, rather than broadly shared wages and incomes, the effective tax base may fail to keep pace. Freeland’s concern about international tax co-operation may be relevant to that problem. But it is a separate argument about distribution and tax design, not a consequence of Baumol’s disease.

Nor can Baumol’s mechanism by itself explain why services should become worse. With the same public-sector workforce and unchanged productivity, the model delivers the same services at an unchanged tax share. If taxes are taking a larger share of GDP while the quantity or quality of services is deteriorating, something else must have changed.

Baumol’s insight explains why a teacher’s, nurse’s or musician’s time becomes more expensive relative to manufactured goods. It does not show that a richer society must surrender an ever-growing share of its income to provide the same amount of that time. A productive economy can afford unchanged labour-intensive public services perfectly well. What it cannot do is make them progressively cheaper relative to everything else.