May 1, 2023
We have constructed a simple input-based Cobb-Douglas type production function model for most countries globally.1 This model enables us to estimate the change in Total Factor Productivity (TFP) on a residual basis, which measures the efficiency of growth input usage, i.e., capital and labor, over time. Utilizing the dataset from the latest IMF WEO, we can also deduce the TFP growth that the IMF implicitly assumes in its projections for the years 2023-2028.
An interesting takeaway from this exercise is that the IMF appears to be optimistic about productivity (TFP) growth. As the following chart demonstrates, the IMF's headline GDP growth forecasts implicitly assume an acceleration in TFP growth compared to the pre-COVID decade's average in most countries for which we could construct these numbers. The IMF only publishes headline GDP growth, so these numbers are "reverse engineered" from Penn World Table (PWT) data and some reasonable assumptions.2
This perceived IMF optimism regarding TFP growth may result from several factors:
1. The IMF may genuinely anticipate a global acceleration in TFP growth. Artificial Intelligence (AI) could be a game-changer, for instance, or biotech advancements could boost some countries (e.g., recent advances in malaria vaccination). This contrasts with a recent World Bank paper arguing that productivity growth will be weak in the near term without comprehensive structural reforms, mentioning the risk of a "Lost Decade" in terms of economic growth).
2. Another reason for this apparent acceleration in TFP growth might be that the IMF is optimistic about working-age population/employment growth rather than TFP growth. In our calculations, we allowed previous employment ratio improvement trends to continue and assumed a stable employment ratio in other cases, but the apparent TFP acceleration remains. The UN population projections we used (see chart below) assume low net migration compared to pre-COVID averages. For some developed countries, working-age population growth could be stronger than UN forecasts if immigration returns to pre-COVID trends, despite political resistance, in countries where the working-age population would otherwise be shrinking. However, this can only significantly affect a limited number of countries, with the flip side being lower working-age population growth in countries with higher emigration. It is likely that working-age population growth will slow in most countries, with many experiencing declines, especially in middle- and high-income countries. This will create pressure to allow more immigration to these countries and/or employ more labor-saving technologies.
3. An additional factor could be that we already have some negative shocks in the past data (e.g., wars and domestic conflicts), which are difficult to forecast and are thus not likely "assumed" in the IMF projections.
4. Countries with large commodity sectors are not well-suited to factors of production analysis. In these countries, governments or non-residents typically have substantial revenues from commodity production. While non-residents often repatriate their profits, governments may not invest to increase production further but instead spend on public goods (e.g., infrastructure). These investments may increase well-being but may not initially boost apparent GDP. Some of the apparent lowest TFP numbers in the past belonged to commodity exporters like Kuwait or Azerbaijan.
5. In theory, it is also possible that the IMF is positively biased towards countries where it has outstanding loans, but there is no sign of such optimism in the estimated TFPs. The IMF expects a very similar median acceleration in TFP growth in countries with IMF programs and those without (1.1 percentage points).
For an individual country, optimism about TFP growth can have a significant impact, but the effect on aggregate global growth is minimal, around 0.2 percentage points. The reason is that while the IMF is optimistic about TFP growth in most countries, it expects a slowdown in TFP growth in the largest economy (in PPP terms), China.
In conclusion, the IMF's implicit optimism about productivity growth may stem from various factors, such as technological advancements, changes in working-age population growth, past negative shocks, and the unique characteristics of countries with large commodity sectors. While this optimism can have a significant impact on individual countries, its effect on global growth remains small due to the anticipated slowdown in TFP growth in China.
1 Methodology: We used Penn World Table data for capital stock, employment, labor share and capital depreciation rate, IMF data for investment ratios (for calculating future capital stock), UN 2022 population projections for working age population. We calculated employment ratios (employed/15-64 aged population) and allowed an improvement in the employment ratio if there was a pre-covid improvement trend. We projected a stable employment ratio in all other cases, thus implicitly have an optimistic forecast about future employment trends. We then calculated a simple TFP as a residual from these variables. The Penn World Table does something more complicated, they estimate human capital improvement and in some cases hours worked as well, so the results are not comparable to their TFP numbers. Because we used a mix of data sources, some caution is warranted interpreting individual countries’ results, but the overall pattern of an acceleration in assumed TFP growth across most countries should be more robust to assumptions, data sources used etc.
2 This calculation does not consider capital capacity utilization, which was likely relatively low in many countries in 2009 and comparatively high in 2019. As a result, excluding capital capacity utilization introduces a bias in the results towards an apparent TFP slowdown rather than acceleration. There may be some implicit IMF assumptions about the ongoing recovery of capacity utilization in the coming years due to the lingering effects of the COVID-19 pandemic; however, this process should be completed by 2028. When creating a chart that only displays implicit TFP growth for 2028, the results remain strikingly similar (not shown here).