Another look at Swedish GDP per capita growth
Increased productivity growth and returns to education are necessary to counteract the ageing population's drag on GDP per capita growth.
Spoiler: Our GDP per capita growth is declining. It is declining due to declining contributions from productivity growth and human capital. Capital per output does not matter much.
The declining contribution from human capital is also a consequnce of higher prosperity. As incomes have grown, we have also chosen to work less and average hours per employee had decreased over time. We have also chosen to have fewer children and smaller families. That has over time led to fewer people in the working age population.
Somewhat counteracting these developments are postivit contributions to GDP per capita growth from the labour force participation rate and education. While the former of these is increasing, the latter is decreasing. This is worrying but it is also something we can address by policies.
GDP per capita can be shown to be a function of the capital-output ratio, human capital per capita and productivity, as shown here. Growth in GDP per capita, gy, is then the sum of the three factors.
Productivity growth is what is left to explain of GDP per capita growth after we have taken into account things that we can measure. So, it can be implementation of new technology, better use of existing technology, more flexible organisations, better management and whatever. Since growth theory predicts that the capital-output ratio is constant, contributions from capital-output to GDP per capita should be small, at least in the long run (see the last graph in this post). Thus, growth of human capital per capita should account for the largest contribution to GDP per capita growth. I hope it doesn’t look like I’m suggesting that there are any casual relationships. This is growth accounting. Accounting explains nothing.
Anyway, below I will study the changes of the contribution of human capital over time. I divide the index of GDP per capita by the index of productivity and graph the resulting ratio. When the ratio increases, the contribution from productivity decreases, and since the contribution from the capital-output ratio is small, the contribution from human capital per capita increases. When the ratio decreases, the reverse happens.
Human capital and productivity matters. Physical capital, not so much.
The ratio grows from 1960 until the end of the 1970s when the Swedish economy experienced crises caused by irresponsible fiscal policies and wage inflation. Several devaluations did not solve this and the fixed exchange rate had to be abondoned during our home made financial crisis 1992. After a few years after the abolishment of the fixed exchange rate, the ratio declines and levels off, indicating that the contribution from human capital per capita is stagnating.
GDP per capita relative productivity.
Source: AMECO database, European Commission. Penn World Tables, Groningen Growth and Development Centre.
Another, and longer, look at human capital per capita.
Those of you that have read my ramblings before, may have noticed that I've blogged about GDP per capita growth before. I have done something similar here, here, and here.
I’m at it again because I read three posts by Dietrich Vollrath, whose book Fully Grown inspired me to previous posts. Anyway, I noticed that he used a measure of the human capital component, education from the Penn World Table. In one of my previos posts, I tried to construct my own mesure of education. Unfortunately, I only had access to data from 1985 so my growh accounting period was very short. If I combine that with data from AMECO, I can extend the time period backwards. The data for education ends in 2019. Here is a description of how the education data from PWT are constructed.
The specification of GDP per capita growth is almost the same as Vollrath’s, which you can find below the heading “Technical Stuff” here. While Vollrath has four components of human capital here, I have five components. My first and second are the same as his, education from PWT, and hours per worker. My third component is the employment rate which I define it as the numbers of workers relative to the number of people in the labour force. The component which I name the labour force rate is defined as the number of people in the labour force relative to the number of people in the working ages. My last component is the same as his, the number of people in the working ages relative to the total populaton. I have used this specification before, in this post about real labour earnings.
Anyway, the new data have not changed my conclusions in my previous posts. GDP per capita growth is slowing down, the contributions to GDP per capita growth from the capital-output ratio, human capital per capita and TFP (productivity growth) is slowing down. Decomposing human capital per capita growth shows that only the contributions from education and labour force participation rate are contributing positively. The contributions from those factors more than outweigh the negative contributions from average hours per worker, the employment rate and the working age population rates.
Average annual GDP per capita growth 1960-2019 (%)
Source: AMECO database, European Commission. Penn World Tables, Groningen Growth and Development Centre.
However, while the contributions from education is positive, its growth rate is declining. The contribution from the labour force participation rate is the only factor which keeps increasing.
This shows up in the ten year growth rates below where we also can see that the employment rate fluctuates just below zero, that the working age population rate is declining and that hours per worker is stagnating. The large negative growth rates of hours per worker in the beginning are mostly due to changes in legislation. In 1970, the working week was reduced from 48 to 40 hours. After the Financial Crisis, it seems that the substitution effect dominates as average hours increase slightly until the two effects cancel out. For the whole period though, the income effect dominates.
Ten year growth rates of human capital components.
Source: AMECO database, European Commission. Penn World Tables, Groningen Growth and Development Centre.
Put more educated people to work.
I haven’t reached any new conclusions in this post. The old conclusions from this and this post are still valid. The ageing problem is difficult to address at least in the short run. As I wrote previously, in the first of the two posts linked to above,
The positive effect from the labour force participation rate is a consequence of the labour force growing faster than the population between 15-64 years old, the working age population. And the working age population rate keeps decreasing. The reason behind the decrease is a consequence of our choices. With rising incomes, we have chosen to have fewer children which after some time translates into fewer people entering the working age.
Thus, we must do something to increase the return of education if the trend of declining human capital growth is to be reversed or at least halted. Because, as I wrote in the second of the two posts linked to above,
The return of education is lower in Sweden than in many other countries. The wages are more compressed, and taxes are higher. Swedish students are older than students in most OECD countries. The average age of a new student was almost years compared to the OECD average of 22 years. This means that the average Swedish student enters the labour force later than in most other OECD countries. And, as Matt Clancy shows, the age of scientists is negatively associated with their productivity, the relatively higher age of Swedish scientists adds negatively to the human capital stock.
The good news is that this can be done faster than reversing the demographics. And it can be done by policies. Let’s do it!
Finally, the capital-output ratio
As mentioned above, here is a graph showing the implied constancy of the capital-output ratio. It moves slightly around zero.
Ten-year growth rates of the capital-output ratio
Source: AMECO database, European Commission.