However, there is a key way that the technological change embedded in GDP-B could well matter for monetary policy – the way it influences how expenditures take place and how they are shifted through time.

Technological change isn’t just about some aggregate level. It changes the composition of products offered, the way buyers and sellers share risk, the information available to both parties, and the level of competition.

All these things have a huge economic impact that monetary authorities will need to consider.

One example I think is important is the shift from durable goods to rental products – through the rise of subscription services.

Subscriptions. Renting instead of purchasing durables.

Instead of buying an album from the most recent Disney movie I can sign up to a subscription to Spotify. By doing so I get access to the Disney music for a limited period of time, for a much smaller cost than buying the album.

This is an example of me buying a subscription to purchase a flow of services, rather than paying for a durable product up front that will provide a stream of durable services through time!

The more we switch to buying subscriptions instead of durable goods, the more we are paying the rental for the durable product as our expenditure – rather than spending a large amount on the durable good all at once.

As our expenditure on these products is now smoothed through time, the variation in demand due to the timing of durable good purchases is smoothed out. This helps to remove a key source of variability in the economy.

However, monetary policy partially works by shifting when people spend on these “durable items”. If they are less important, then changes in interest rates will also shift output by less.

The idea here is that if the cycle is “smaller”, the central bank doesn’t need to shift output by as much. However, for the same reason their policy can’t shift output by as much either!

So how does this work?

Imagine everyone stopped buying durable goods, and replaced those by purchasing a flow of digital services . In this case the purchase of the digital product is less lumpy than the corresponding durable good was, and so changes in the time profile of interest rates are less effective at getting you to shift when you spend on it.

For example, a lower interest rate might persuade me to buy a washing machine – but if I just hire a machine, a lower interest rate won’t convince me to pre-pay for all my future uses of the machine. As a result, the change in interest rate is less able to shift around my expenditures.

Worryingly, this also suggests that monetary policy will be less effective for certain shocks through this channel. If there is some other exogenous change in expenditure (eg demand for NZ exports drop), there is less scope to use monetary policy to fill this gap by getting consumers to increase their expenditure on durable products.

Powell touched on three aspects of evolving economy: the consequences of an oil price spike, the measurement of output and productivity, and the role of tightness in the labour market.

In this post I will talk about the measurement of output and productivity aspect of the speech and some elements of that which will matter for monetary policy.

In my earlier post, I have described a new GDP-B metric introduced by Erik Brynjolfsson. Powell’s speech placed special emphasis on the role of that measure, and as a result I am going to talk about the monetary policy consequences stemming from this GDP adjustment.

In summary, there are three important consequences from any technological change:

• How technological change influences the use of factors of production.
• How technological change influences the response of firm investment to changes in expenditure.
• Howe technological change influences the response of consumers – and to the products provided – to changes in overall expenditure.

Given these changes, for monetary policy there are two channels that matter for this response:

• It’s influence on the neutral interest rate
• It’s influence on the responsiveness of output to changes in the interest rate

As a result, we’d like to think about how the technological change embedded in the GDP-B measure would change the first three issues, and thereby influence these two channels.

GDP-B and higher income growth

Brynjolfsson argues that if we incorporated the rise in consumer surplus obtained from free services provided by digital goods, we would end up in higher GDP growth numbers.

Now, how does this relate to the neutral interest rate?

It is easy to imagine that this would imply we need higher interest rates. Imagine a situation, where we had a sudden increase in GDP that we then view as providing a positive output gap. To respond to the shift in GDP, central bank would increase their interest rates.

But is this what is happening? No. GDP-B embeds “technological” change which changes the way factors of production turn into output. As a result, even if this does increase the GDP number it has also shifted “Aggregate Supply” to the right as well.

Our new measure of GDP does not tell us that it is additional “Aggregate Demand” or “Aggregate Supply” that is involved in the higher observed GDP number – all it tells us is that incomes are being undermeasured in a sense. It tells us nothing directly about the output gap.

Instead we need to rely on prices.

GDP-B and prices

GDP-B makes it clear that digital provision is leading to lower prices, or even free goods, which are increasing consumer surplus. Do these falling prices suggest lower interest rates?

At first glance no. A change in relative prices due to the provision of products in GDP-B does not indicate interest rates need to be lower – as a central bank will look through this price change (inflation expectations are expected to be unchanged).

But what happens if the rate of declining prices is expected to persist. Specifically, what happens if the importance of non-rival and free goods is expected to grow over time?

There are two things in this case:

• Lower average price growth in the future will be expected than in the absence of the change, implying that we need lower interest rates now to keep inflation expectations the same.
• If this is representative of a future where there are additional profitable investment opportunities from these future technologies (eg highly scalable business models) then this could increase current investment. This will drive up demand now, even though the factors of production now haven’t changed. This would suggest interest rates need to be higher now.

Ultimately, understanding which behaviour is most relevant will have a big impact on what we think is happening with the neutral interest rate. And our best indicator of all this isn’t a GDP-B measure – but appropriate measures of prices and inflation expectations.

Does looking at prices tell us everything we need to know about technological change and monetary policy? Not quite. Next week I will discuss how technological change may influence how responsive the economy is to monetary policy.

The focus of this post is on digitization. In an era of digitization, economists have become more and more concerned about whether the conventional way of calculating GDP is appropriate for asking questions about changes in consumer welfare (surplus) through time.

GDP associates a dollar value with a product in terms of its market price and aggregates up all the products – we can think of this as a proxy for the way society could trade-off between the products, and also the relative value consumers of the final good have over the products they are buying. In this way, an increase in GDP is associated with an increase in consumer surplus in traditional goods markets .

With the rise of free services provided by digital goods such as free apps on smartphones, Facebook and etc, Erik Brynjolfsson notes that we can end up in a situation where the GDP can stay constant (or even falls), but the technological change increases consumer surplus. Brynjolfsson proposes a new GDP measure called “GDP-B” where he adds contributions from digital platforms into the GDP growth. 

Why should we make adjustments to the conventional GDP measure?

Imagine a world where we all had a product that gave us a whole bunch of service value through time. The value of the (durable) product would be measured in GDP as a purchase, but the flow of services would not – so when describing the “flow” of consumption there would be a pretty severe problem.  Hence why we put things like “imputed rents” into the GDP data.

Furthermore, we may expect this durable good price to represent the value of the services it provides – but in the face of competition, the price of the durable good may be lower than the price of the services it is replacing.

Now imagine a world where a paid product suddenly becomes “free”.  We know there is a “price” somewhere as the price of the cellphone will now in some way reflect this additional value, but how well does that capture the increase in “consumer surplus” that is occurring.  If we are trying to use GDP as a reflection of welfare this is an issue.

Both Sharon Pells from MBIE and the Productivity Commission have recognised the importance of these measurement issues, and how they may give us a misleading read on consumer surplus.

Tying these points together, the digital economy leads to the provision of a huge number of previously paid services for *free, on the basis of purchasing a cellphone and a cellphone plan. Bundling together all these final services (the things consumer actually value) there has been a massive decrease in the cost of purchasing these services. However, looking at GDP measures we would see a decline in GDP in the industries that previously provided them (eg Kodak cameras) and an increase in purchases of ICT equipment.

If we measured the final consumer service all that has happened is that prices have fallen for market provided services, and so consumer surplus would have risen. But our way of measuring doesn’t capture this reality as it is based on these intermediate product types rather than the final service to the consumer.

Erik Brynjolfsson suggests an adjusted measure, GDP-B, to augment traditional GDP for new and free goods – with the goal of more closely associating changes in the new measure with consumer surplus.

The idea behind these measures is to associate a reservation price with goods that are provided, that is estimated to be the lowest price which would have had zero quantity demanded before the product existed.

1. For “new goods” this gives an indication of the reduction in the price associated with the introduction of the good, and a measure of surplus.
2. For a “free good” it is necessary to calculate their compensating variation, or how much expenditure would they need in order to be indifferent to losing the good.

To estimate these, Brynjolfsson used a series of large scale, cool, experiments.

For instance, he ran an experiment with cellphone camera usage to identify the importance of quality adjustments for goods with rapid quality change. People would state an amount of money they would be willing to receive in order to avoid using the camera, and then would have their camera taped up – if they took off the tape they would forfeit the money, providing a revealed preference of their valuation!

The experiment gave a strong evidence that consumers obtain a significant amount of surplus from using their smartphone cameras.

So this replaces GDP?

Brynjolfsson, Collis and Eggers suggest that we shouldn’t be replacing GDP, but rather use a different metric to capture consumer wellbeing through consumer surplus, which in sense a new GDP-B metric does.  

“We propose a way of directly measuring consumer well-being using massive online choice experiments. We find that digital goods generate a large amount of consumer welfare that is currently not captured in GDP.”

As they note “GDP is a measure of production not wellbeing”. If the question at hand is one regarding domestic production, or the use of factors of production, GDP still has a role. Additional measures are just more appropriate when we have different questions.