Productivity puzzle - due to an overestimating of inflation?

This ½% inflation overestimation reflects the mismeasurement of IT hardware and software prices, new product and new store bias, free information and the sharing economy. Meaning real GDP and living standards are better than commonly assumed, explaining strong jobs growth.

While I can’t (yet) 'set my phaser to stun’, I can get real-time satellite navigation, stream music whenever I want, make free video calls, and get instant universal translation – all for free – on my little pocket device.

Yet according to common folklore, productivity growth is weak. Real incomes haven’t risen. Living standards are being squeezed...you know the story.

I'm a techno-optimist. Robert Gordon is the arch techno-pessimist. And he argues that the computer revolution peaked in the late 1990s.

It certainly did in the GDP data. But that doesn’t mean it’s peaked in the real world.

In the late 1990s, statisticians revised down computer prices, and revised up real GDP, when they realised that $1,000 got you a faster computer with more storage each year.

The same statisticians now say computer prices are no longer falling, acting as a drag on productivity. But this is disputed by Fed researchers Byrne, Oliner and Sichel in their paper "How Fast are Semiconductor Prices Falling?.

There is even bigger uncertainty about software prices. Most statisticians assume little productivity in software and instead deflate nominal software investment by programmers’ salaries. The US uses a combination of programmers’ costs and ‘packaged software’ prices. But again, Fed researchers believe this subcomponent is falling far more rapidly than estimated (see “Seasonality, Consumer Heterogeneity and Price Indexes: The Case of Prepackaged Software", Copeland, 2011).

Although computer prices were ‘quality’ or ‘hedonically’ adjusted in the 1990s, they don’t do this for services. UK telecom services prices are allegedly HIGHER today than in 2000, even though we’ve swapped our text and minutes for data.

Free information is messing the statistics up. We used to pay a commission to a travel agent to find us a nice place to stay. We now get that information for free using the internet. This has saved us money (commission) which we can spend on other things. But the value added from the travel agent is missing from the GDP data, even though we still have it.

Ditto newspapers. Physical sales have halved, but we’re reading more online (for free).

The sharing economy appears under recorded. Initial estimates of GDP are measured by surveying big companies. An alternative measure based on (lagged) tax returns, is running ¼% higher per year. This happened in the UK in the late 1980s when statisticians were focusing too much on manufacturing and not enough on services.

Finally, in the UK we estimate food-price inflation is overestimated by at least 2% per year as a result of the rise of discounters like Aldi. Statisticians only take into account the INDIRECT effects (Tesco cutting some prices to match) but not the DIRECT effect (households switching to cheaper stores).

Adding it all together, we estimate that inflation is probably overestimated by ½% and real GDP/productivity underestimated the same amount.

For a more detailed look please see Fundamentals “Bean Counters”