Tag: productivity

Posted on

The role of capital in agricultural productivity growth

Discussions on the future competitiveness of European agriculture often return to the role of capital. Investment in modern equipment, digital technologies, and precision farming systems is widely regarded as essential to sustain productivity growth and to meet new environmental and market challenges. While the Draghi report on EU competitiveness did not directly address the agriculture and food sector, questions are increasingly asked about whether the current pace and composition of investment are adequate, and whether capital is being used efficiently within the sector. These questions are particularly relevant as structural change continues, with a declining labour force and a growing reliance on technological and digital solutions.

The relationship between capital, labour and output in agriculture provides a useful framework for examining these issues. In accounting terms, labour productivity—measured as output per worker—depends both on the amount of capital each worker uses (capital per worker) and on how efficiently that capital is employed (output per unit of capital). Higher investment in machinery and equipment can raise labour productivity if it enables workers to produce more output, but it can also lead to over-capitalisation if output does not increase in proportion to the additional capital used. The direction and efficiency of capital accumulation therefore matter as much as its scale.

A caveat is that, in agriculture, land remains a central determinant of productivity. The amount and quality of land available per worker influence both labour and capital productivity, often more strongly than reproducible capital itself. While the discussion that follows focuses on the relationships between output, labour and capital, we recognise that part of the observed variation in productivity across countries will reflect underlying differences in land endowments and land use intensity.

Against this background, this post examines how capital is being used within EU agriculture and how it relates to overall productivity performance. I make use of a relatively new Eurostat domain in its national accounts database which calculates productivity indicators. The following section looks at recent trends in three key ratios—output per worker (GVA/L), capital per unit of output (K/GVA), and capital per worker (K/L)—to see what insights can be drawn about the role of capital in shaping the competitiveness and structural evolution of European agriculture. We will see that important data issues prevent drawing robust conclusions.

Trends in net capital stock

We start by examining the growth in the capital stock. Figure 1 shows the change in the net capital stock in chain-linked volume in the sector ‘Agriculture, forestry and fishing’ (AFF) between 2005 and 2022 (see the data appendix at the end of the post for further explanation, the Eurostat data on productivity are not available for the agriculture sector alone but this is a reasonable proxy as agriculture accounts for 87% of the GVA in the AFF sector). Data for some countries are available up to 2024 but not all, so I have restricted the period to 2005 to 2022.

For the EU27 as a whole, there has been almost no change in the volume of the sector’s capital stock. However, the range of country experiences is extraordinary. Countries to the left of the EU column in Figure 1 have been decapitalising, while countries to the right have been adding to their capital stock over this period. If we arbitrarily draw a line between Belgium and Lithuania, the countries from Lithuania and further to the right in Figure 1 where net investment has been particularly marked are all countries that joined the EU in 2004 or after (except for Luxembourg). Latvia and Croatia have more than trebled their capital stock over this period. There are exceptions. Slovenia increased its capital stock at a much slower pace, while Hungary, Poland and Cyprus have seen a decrease in their capital stock. What explains these very different experiences with respect to capital formation deserves a separate investigation.

Figure 1. Percentage change in net capital stock in the Agriculture, Forestry and Fishing sector between 2005 and 2022.
Source:  Eurostat, Capital stocks by industry (NACE Rev.2) and detailed asset type [nama_10_nfa_st], chain-linked volumes, 2020 prices.

Trends in gross value added

As background to help interpret some of the productivity ratios later, Figure 2 shows the growth in the volume of GVA in the AFF sector over the period from 2005 to 2022. GVA can be volatile from year to year, suggesting caution in looking at changes between single years. Thus, for Figure 2, I have constructed the percentage changes using 3-year averages at the start and end of the period.

Total growth in the EU27 was a rather unimpressive 10% over the 17 year period, but again with very considerable heterogeneity. In one-third of Member States (9/27), the sector contracted, sometimes by considerable amounts (-31% in Cyprus, -12% in Greece, -5% in Italy and -3% in Denmark). At the other extreme, AFF GVA is reported to have more or less doubled over the same period in Ireland and Slovakia.

As I was puzzled over some of the numbers shown, I have added in Figure 2 the growth in the volume of agricultural GVA over the same period based on the Eurostat Economic Accounts for Agriculture (EEA).  Some difference is to be expected as the National Accounts data cover the sector ‘Agriculture, forestry and fishing’ while the EEA data refer only to agriculture (the national accounts data also provide the series for agriculture alone and comparing this series to the EAA series shows similar differences). To calculate the volume of GVA, both series use chain-linked volumes and the double deflation method (see data appendix at the end of this post). The national accounts data are chain-linked using 2020 as the reference year, while the EEA data are chain-linked using 2015 as the reference year, which could account for some differences.

While the overall growth in the volume of EU GVA and for some countries is reasonably similar in the two series, there are striking differences for most countries. The most dramatic difference is for Slovakia, which reports almost a doubling (86% increase) in its GVA volume in the National Accounts but a 31% decrease in the EEA! Other countries which show significant divergences include Lithuania (36% as against 157%), Denmark (-3% as against 61%), Ireland (108% as against 53%), Lativa (30% as against 90%), Finland (26% as against 63%), and Poland (1% as against 45%). When we observe such extraordinary differences for estimates of the growth in output for two very similar series, it calls into question how to interpret any productivity growth figures based on these estimates. It is clear the reason for these differences deserves interrogation.

Figure 2. Percentage change in volume of GVA in the Agriculture, Forestry and Fishing sector (National Accounts) and the Agriculture sector (Economic Accounts for Agriculture) between 2005 and 2022 using three-year averages (2005-2007 to 2020-2022).
Source: Eurostat, Gross value added and income by main industry (NACE Rev.2 ) [nama_10_a10]: Economic accounts for agriculture – indices (volume, price, values) [aact_eaa05], chain linked volumes, index 2020=100.

Capital productivity and capital intensity of production

Despite these concerns over the interpretation of data, we rely on the Eurostat productivity data using the national accounts figures for the Agriculture, Forestry and Fishing sector in the remainder of this post. Figure 3 shows the capital intensity of AFF production, as measured by the ratio of capital (net capital stock in chain-linked volumes) to AFF output in volume terms (gross value added in chain-linked volumes). Its inverse, the ratio of output to capital, is a measure of capital productivity. We focus on changes in the capital intensity of production by Member State over the period 2005-2022 in Figure 3. For this and subsequent figures, I use the percentage changes between the single years 2025 and 2022.

Figure 3. Percentage change in the capital intensity of output in the Agriculture, Forestry and Fishing sector between 2005 and 2022.
Source: Eurostat, Capital stock-based productivity indicators by industry (NACE Rev.2) [nama_10_cp_a21], index, 2020=10.

Overall, the capital intensity of EU agriculture appears to have fallen slightly, driven by the falls in capital intensity (or, putting it otherwise, increases in capital productivity) in the countries to the left of the Figure. Capital intensity is a ratio, so its trend is influenced both by the growth in the stock of capital (Figure 1) and growth in the volume of output (Figure 2). The big reductions in capital intensity in Slovakia and Ireland are driven by the sharp rise in the volume of output reported in the national accounts, as the capital stock also increased in both countries.

More plausible is the significant increase in the capital intensity of agricultural production in the countries to the right of the figure, which are all countries (except for Luxembourg which has added significantly to its capital stock in the last two decades) that acceded to the EU in 2004 and after. Poland is an important exception where capital intensity has decreased due to a small increase in output accompanied by a slightly greater fall in the capital stock.

Labour productivity

Analogous to capital productivity, we can also calculate a measure of labour productivity, defined as the output (measured as gross value added) in the AFF sector divided by the number of hours worked in the sector (see data appendix at the end of this post for a justification). Figure 4 shows the growth in labour productivity in Member States over the period 2005-2022 and, again, huge heterogeneity is evident. In four countries, the growth in labour productivity has been negative. Each hour spent at work in farming in these countries contributed less in 2022 than it did in 2005, which is a remarkable finding. There are another three countries – Italy, Belgium and Sweden – where the percentage growth over the period is less than 20%, or less than 1% per annum. Those countries to the right of the graph show above-average increases in labour productivity. There is a much less distinctive pattern in the growth of labour productivity between old and new Member States in Figure 4. While new Member States are to be found to the right of the chart, so are countries such as Ireland, Portugal, Finland and Germany, while countries such as Malta, Cyprus, Czechia, Slovenia, Poland and Bulgaria have below EU-average growth in labour productivity.

Figure 4. Percentage change in labour productivity in the Agriculture, Forestry and Fishing sector between 2005 and 2022
Source: Eurostat, Labour productivity and unit labour costs by industry (NACE Rev.2) [nama_10_lp_a21]. Real labour productivity per hour worked, index, 2020=100.

As for capital productivity, a change in labour productivity is a ratio which is influenced both by the growth in output (Figure 2) and the change in labour input (measured in hours worked) (Figure 5). The dominant trend is towards a reduction in labour input, but there are three countries – Belgium, Luxembourg and Malta – where labour input increased between 2005 and 2022.  There are six countries, all in Central Europe except for Portugal, where labour input fell by 47% or more.  The fall in labour input is clearly a major contributor to the increase in labour productivity in the AFF sector shown in Figure 4.

Figure 5. Percentage change in labour input (hours worked) in the Agriculture, Forestry and Fishing sector between 2005 and 2022
Source: Eurostat, Employment by main industry (NACE Rev.2) – national accounts [nama_10_a10_e]. Total employment domestic concept, index 2015=100 (based on hours worked).

Capital deepening – the capital to labour ratio

Another possible contributor to an increase in labour productivity is an increase in the capital stock per worker. A positive change in the capital-labour ratio is referred to as ‘capital deepening’ and can be an important contributor to increased labour productivity, though more so in industry where land is not as an important factor of production as in agriculture. Trends in the capital-labour ratio are shown in Figure 6.

Figure 6. Percentage change in the capital to labour ratio (net fixed assets per hour worked) in the Agriculture, Forestry and Fishing sector between 2005 and 2022.
Source: Eurostat, Capital stock-based productivity indicators by industry (NACE Rev.2) [nama_10_cp_a21]. Net fixed assets per hour worked, index 2020=100.

The general tendency is for workers in the AFF sector to have more capital to work with over time, though Cyprus and Italy are two exceptions. But for the six countries to the right of the graph, all acceding countries to the EU in 2004 and after, the increases are quite dramatic (note the scale on the vertical axis). This reflects both a significant increase in the capital stock (Figure 1) and also a bigger-than-average decrease in labour input (Figure 5). Where capital deepening is a consequence of the outflow of labour, it does not necessarily represent productivity-driven investment. However, as noted, many of these countries have also been adding to their capital stock which is more likely to enhance the productivity of labour.

Relationships between the productivity indicators

Relationship between GVA/L and K/L. To further investigate the possible relationship between increases in the capital-labour ratio and increases in labour productivity in the AFF sector, Figure 7 shows the correlation between the variables. Overall, the figure shows a very weak correlation. Observation suggests that the slightly positive slope is heavily influenced by the Lithuania outlier and, indeed, if this is removed, the trend line becomes almost flat. There is also very large variation around the trend line (captured by the low R2 value of 0.16). That is, countries that experienced similar rates of increase in the capital-labour ratio also had very different growth rates in labour productivity.

Figure 7. Correlation between the percentage change in labour productivity and change in the capital-labour ratio in the Agriculture, Forestry and Fishing sector between 2005 and 2022.
Source:  Own construction, based on Eurostat national accounts data described in previous figures.

There is thus limited evidence that more rapid growth in capital per worker is associated with more rapid growth in labour productivity in the AFF sector. This could point to a poor allocation of the additional capital, leading to possible overcapitalisation or poor capital utilisation. It also reflects the fact that labour productivity in the agricultural sector is more influenced by the availability of land per worker, and a more thorough analysis would need to investigate the impact of additional capital per worker holding land endowment constant.  

Relationship between K/GVA and K/L. It is also informative to look at the relationship between the capital intensity of production (recall that this is the inverse of the productivity of capital) and the capital-labour ratio (Figure 8). The trend line here has a stronger upward slope with less variation around the trend. The line indicates that countries with a faster rate of increase in the capital-labour ratio also had a faster increase in the capital intensity of production (or a faster decrease in the productivity of capital). This supports the view that there may be diminishing returns as farmers add more capital per worker. However, the relatively low correlation coefficient indicates that other factors are also important in explaining this relationship.

Figure 8. Correlation between the percentage change in the capital intensity of production and change in the capital-labour ratio in the Agriculture, Forestry and Fishing sector between 2005 and 2022.
Source:  Own construction, based on Eurostat national accounts data described in previous figures.

Further insights are gained if we mentally divide the X-Y graph into four quadrants at the (0,0) point. In the bottom right-hand quadrant are those countries (mostly the older Member States) where the capital-labour ratio has been increasing but the capital intensity of production has been decreasing (capital productivity increasing). These are countries where capital deepening has been efficient in that output per unit of capital has been maintained or improved.

Countries in the upper-right quadrant have experienced both rising capital-output and capital-labour ratios. So there is substitution of capital for labour but without clear productivity gains, possibly reflecting diminishing marginal returns to capital or capital misallocation.

A major caveat to all the trends discussed in this post is the significant discrepancy in output growth measurements between two official Eurostat sources (see Figure 2). Until these discrepancies are resolved or explained, the specific magnitudes of productivity changes for several Member States, and thus the strength of the relationships drawn, should be interpreted with caution. The overall direction of the trends for the EU27, however, is likely robust.

Conclusions

This post has examined simple productivity trends in the Agriculture, Forestry and Fishing sector using Eurostat national accounts data over a period of nearly two decades, with a focus on the role of capital. It makes use of the accounting identity that labour productivity, measured as output per worker, depends both on the amount of capital each worker uses (capital per worker) and how efficiently that capital is employed (output per unit of capital).

Data on the change in the capital stock in agriculture show disparate trends. In one-third of EU Member States there is a process of decapitalisation where new investment is less than the depreciation of the previous capital stock. This includes countries like France, Germany and Denmark. It is hard to match these data with stories that farmers are investing in ever-bigger machines, in robotics and in new stables or slurry tanks to meet higher animal welfare or environmental requirements. On the other hand, the data show that many of the new Member States have undertaken massive new investments and have significantly increased their capital stock, but with some evidence that this does not always translate into higher productivity.

We observe a great heterogeneity of growth rates both in the growth of labour productivity over time, growth in the capital-labour ratio and growth in the capital-output ratio.  With respect to the growth in labour productivity, there is no real hard and fast distinction between old and new Member States, which will surprise many (it surprised me!). In fact, in four countries labour productivity has fallen over the past two decades.

Labour productivity is an important determinant of income from agriculture (together with subsidy transfers and outgoings for non-owned production factors of land and capital). One would have expected that the new Member States, with lower average levels of labour productivity to start with, would be converging on labour productivity levels in the older Member States through more rapid growth. But this expectation does not appear to be supported by these data.

We also conclude that more rapid increases in the ratio of capital to labour are not strongly reflected in more rapid increases in labour productivity. While this may point to inefficiencies in capital allocation, the most plausible explanation is the unobserved role of changes in land per worker and land quality. A more robust analysis would require a multi-factor productivity model that explicitly includes land.

One question is whether it is reasonable to assume that the trends in the AFF sector can be taken as representative for the agricultural sector alone. The use of the AFF sector is because we only have capital stock data for this sector and not for its individual components. But agriculture consistently accounts for about 87% of the GVA in the AFF sector, so there should be a close correlation between movements in the two series.

The bigger question which throws a shadow over all the subsequent analysis is whether we can take the data used in this post at face value, given the dramatic differences revealed in the growth in the volume of output between the national accounts and the economic accounts for agriculture (while one series in Figure 2 measures output in the AFF sector to be consistent with the capital stock measure, and the other output in agriculture alone, the same differences exist if we make the comparison with just the agriculture sector alone in the national accounts data). I am not aware of anyone who has put these two series together before now, so I cannot offer any explanation for these discrepancies. Anyone who has insights is very welcome to share them in the comments.

Data appendix

The data I use is mostly drawn from Eurostat’s labour and capital productivity domain nama_10_lpc described in detail in these metadata. The data are part of an extended dataset that has been published since December 2021 following the work of a Task Force on productivity. The data are derived from the national accounts provided by Member States and are broken down by economic activity. Data are not available for agriculture alone but only for the larger ‘Agriculture, forestry and fishing’ (AFF) sector. Differences in the composition of this sector between EU countries would be expected to bias a comparison of productivity levels in absolute terms. In this note, I am looking only at trends in productivity over time, which should minimise the impact of such composition differences.

We examine the relationship between three variables:

  • Gross value added (GVA) in the AFF sector, which is the difference between output at basic prices and intermediate consumption. Note that subsidies not related to production, such as decoupled area payments or environmental or rural subsidies under the CAP, are not included in GVA. GVA is a measure of the economic output generated by the sector itself. GVA is measured using the double deflation method in chain-linked volumes (GVA in prices of the preceding year is expressed as the difference between output measured in prices of the preceding year and intermediate consumption measured in prices of the preceding year), so it is an indicator of the real level of economic activity excluding the impact of inflation.
  • Labour input (L) includes both employees and self-employed persons and can be measured either in terms of persons employed or hours worked. The methodology is consistent with the other national accounts variables (for example, the employment figures must be consistent with the compensation of employees data). It thus differs from the employment estimates published by the Labour Force Survey or the Integrated Farm Statistics. The purpose of the national accounts employment estimates is to measure the input of labour to the process of production. Individuals with multiple jobs are recorded only under the industry of their primary activity to prevent double-counting but this can create distortions when looking at productivity trends if the share of part-time workers changes over time. For this reason, Eurostat recommends that productivity measures involving labour should used hours worked rather than persons employed.
  • Capital inputs (K) are measured using data on the net capital stock. The net capital stock is the sum of the written-down values of all the fixed assets still in use. It can also be described as the difference between gross capital stock and consumption of fixed capital. Ideally, capital productivity indicators should be calculated using the flow of capital services provided by the productive capital stock, which can be considered analogous to the hours worked measure for labour services. Deriving such a measure requires sophisticated calculations and assumptions and is not currently required of Member States, so the net capital stock is used to approximate this concept. The assumption is that the change in the market value of net capital stocks approximate changes in productive capacity over time (Eurostat). For the purpose of measuring productivity, net capital stocks are calculated in real terms as chain-linked volumes.

This post was written by Alan Matthews

Photo credit: Corn harvest, WFranz via Pixabay and used under a Pixabay licence.

Posted on

What is happening to EU agricultural productivity growth?

Agricultural growth can come about through bringing new resources into production (new land, extension of irrigation, or input intensification per hectare) or through raising the productivity of existing resources. The appropriate measure of productivity growth in this context is Total Factor Productivity (TFP) growth, which is defined as the aggregate quantity of outputs produced by the agricultural sector divided by the aggregate quantity of inputs used to produce those outputs.

This measure contrasts with partial productivity measures such as growth in labour productivity or land productivity (yields per hectare) because an increase in these partial productivity measures can be achieved by increasing the intensity of use of other inputs (for example, crop yields can be increased by applying greater amounts of fertiliser or using more labour). Thus, partial productivity indicators do not measure ‘true’ productivity growth.

A higher level of TFP in agriculture can mean achieving greater output for the same inputs or using fewer inputs for the same level of output. Given the growth in global food demand and the need to minimise the additional demands on resources such as land and water, as well as intermediate inputs such as fertilisers and chemicals for environmental reasons, increasing the growth of TFP in agriculture should be a policy priority.

Indeed, the first of the five objectives for the Common Agricultural Policy set out in the Treaty of Rome and incorporated unchanged into the latest Lisbon Treaty is “to increase agricultural productivity by promoting technical progress and by ensuring the rational development of agricultural production and the optimum utilisation of the factors of production, in particular labour”.

Measuring TFP in EU agriculture

However, despite its policy importance, very little is known about TFP developments in European agriculture. In the early 2000s, Eurostat initiated an effort to develop a Multi-Factor Productivity (MFP) index for agriculture based on the Economic Accounts for Agriculture (EAA). The MFP index is similar to the TFP in that it compares the growth in agricultural output to the growth in a bundle, but not all, agricultural inputs. In particular, the Eurostat index did not take account of changes in land use for computational reasons.

The Eurostat index was published for a couple of years in the early 2000s (it was included in the annual Income from Agricultural Activity reports, for further details see this paper) but was then discontinued.

The need for a TFP index has become topical again following the 2013 CAP reform which puts greater emphasis on the need for monitoring and evaluating the effects of agricultural policy, including the new European Innovation Partnership for Agricultural Productivity and Sustainability (EIP-Ag). One of the indicators to be used to measure the impact of EIP-Ag is the growth in agricultural TFP, and it is included in the set of impact indicators being prepared by DG Agri where the new indicator is described.

The calculation of TFP growth faces many difficulties in terms of conceptual and methodological issues and data availability (for a discussion of these issues, see my earlier post on changes in global agricultural productivity growth). This is illustrated by comparing the preliminary results from DG AGRI’s computations (taken from a presentation by Tassos Haniotis at an IATRC symposium on agricultural productivity last year) with data from a newly-released USDA database on international agricultural productivity growth which also computes TFP measures for EU countries (I am grateful to David Blandford of Penn State University for drawing my attention to this database).

DG AGRI’s TFP measure

We first look at the DG AGRI findings on TFP growth in EU agriculture. As an approximate rule of thumb, the normal expectation of TFP growth in agriculture is around 2% per annum. The first chart below shows that from 1995 until about 2002, TFP growth in the EU-15 was around 1.6% per annum. However, since then EU-15 TFP growth in agriculture has stagnated, growing by only around 0.3% per annum over the period 2002 to 2011.

The only bright spot has been TFP growth in the new member states, which averaged around 1.6% growth per annum over the 2002 to 2011 period. However, these countries account for a relatively minor share of total agricultural output in the EU, so TFP growth in the EU-27 over the past decade was a disappointing 0.6% per annum. Note that to smooth out year-to-year variations in TFP because of yield fluctuations due to weather, the data are smoothed by using three-year averages, so the 2011 data is an average for the years 2009, 2010 and 2011.



TFP growth by individual country is shown in the second chart taken from the Haniotis presentation. This shows the impressive productivity performance of some of the new member states on the left-hand side (the five countries with the highest TFP growth over the 2001 to 2010 period (here I take the mid-year of the three year average rather than the end-year to make clear the similarity of the periods shown with the USDA data below) are all new member states, shown with yellow bars. Finland, Austria, Luxembourg and Denmark are the countries from the old EU-15 member states with the best TFP performance, with Spain, Ireland and Italy showing negative TFP growth over this period.



EU agricultural TFP growth according to USDA data

The other source of information on EU agricultural productivity growth is the USDA International Agricultural Productivity database (Incidentally, these are also the data underlying the Global Harvest Initiative’s annual GAP® report discussed in my earlier post).

The purpose of this database is to provide agricultural productivity growth rates on a global basis for as wide a number of countries as possible, so a comparable methodology based on FAOSTAT data is used. The approach is, in principle, the same as that used by DG AGRI, but the results could not be more different.

According to the USDA data, agricultural TFP growth in the EU-23 (for lack of long-term consistent data, the three Baltic EU countries and the countries of the former Yugoslavia Slovenia and Croatia are omitted, while the Czech Republic and Slovakia are grouped together) has actually been accelerating over the past decade. Furthermore, this acceleration in agricultural TFP growth has been particularly pronounced in the EU-15, while TFP growth in the new member states has slowed down more recently.

Specifically, agricultural TFP growth rates for the EU-23 according to the USDA estimates were 2.1% for the decade 1991-2000, 2.2% for the period 2001-5 and 3.1% for the period 2006-2010 (these figures compare to 1.8% in the 1970s and 1.5% in the 1980s). I have constructed all the averages quoted here as weighted averages where each individual country’s TFP growth rate in the group has been weighted by its average share of the group’s total agricultural output value in current prices taken from Eurostat for the years 1999-2001).

The corresponding figures for the EU-15 were 2.2% for the decade 1991-2000, 2.3% for the period 2001-5 and a whopping 3.5% for the period 2006-2010 (these figures compare to 2.0% in the 1970s and 1.6% in the 1980s).

The new member states show a different pattern. The corresponding figures for the EU-8 were 1.0% for the decade 1991-2000, 1.2% for the period 2001-5 and a disappointing 0.5% for the period 2006-2010 (these figures compare to 0.7% in the 1970s and 0.7% in the 1980s).

Thus, according to the USDA figures, productivity growth in the new member states has been consistently lower than in the old member states, and the gap has grown much bigger in the most recent period. The figure below shows the data for individual member states for the same decade as for the DG AGRI chart above. There is almost a perfect inverse correlation with the DG AGRI data.



Over the decade of the 2000s, Italy, Portugal, Netherlands, Germany, Spain and Austria all had TFP growth of 3% or more, according to the USDA data. Indeed, during the second half of the period, TFP growth in Italy, Portugal and Netherlands reached 5% and very close to that in Austria. Output growth in these calculations has been smoothed statistically to reduce the impact of weather-related yield variation from year to year, so these calculated high rates of TFP growth are very impressive. The disappointing performance of TFP growth in the new member states emerges clearly from the chart.

Why is our knowledge of productivity growth in EU agriculture so uncertain?

Without much more detailed analysis, it is hard to know why these two different sources tell such very different stories about agricultural productivity growth in Europe. Both use information on the growth of agricultural outputs and inputs over the 2001-2010 period which are then weighted to form the aggregate total output and total input indices from which TFP growth is derived as the change in the ratio over time.

The differences can be due to differences in the volume measures for the individual outputs and inputs, to differences in the weights used for aggregation, and to differences in the index number methodology adopted to create the TFP index.

The USDA figures build on FAOSTAT data for outputs and inputs while the DG AGRI figures build on the Eurostat EAA accounts. The latter are much more detailed than those available in FAOSTAT particularly for input use. The weights used for aggregation are in principle value shares. These shares are notoriously difficult to calculate particularly for the farmer-supplied inputs of own labour, capital and land. But because the trends in these inputs are vastly different in the EU (for example, a dramatic fall in farm labour compared to steady increases in capital input), the weights attached to these respective inputs will greatly influence the final TFP estimate. The weights are also influenced by the index number methodology used.

The stories told by these different indicators of TFP growth in EU agriculture are so different that it will be very important to investigate the reasons why these differences arise. The USDA database is highly transparent which should facilitate such a comparison. I understand that DG AGRI plan to release details of their TFP estimates later this year which may allow the reasons for these differences to be more fully assessed.

This post was written by Alan Matthews.

Posted on

Impact of CAP subsidies on productivity

I recently had an exchange on Twitter with Martin Crowe, an Irish dairy farmer and agri-consultant, over the apparent stagnation in Irish agricultural output over the past 20 years (follow on @xAlan_Matthews and @martincrowe). I attributed this, in part, to the role that direct payments play in Irish farm incomes. I argued that “If 70% of your income is coming as a cheque in post there is less incentive to innovate to grow the remaining 30%” (direct payments make up around 70% of Irish farm income in an average year). Martin tweeted back that the “70% gives the security and confidence to try and improve the 30%”.
At issue here is the impact of CAP direct payments on farm productivity. As the Twitter exchange indicates, there are potentially both positive and negative effects.
How direct payments might affect productivity

In the agricultural economics literature, the positive effects rely on credit constraints and assumptions about risk behaviour in agriculture. If farms find it difficult to access credit, then subsidies may provide an additional source of financing, either directly by increasing farms’ financial resources or indirectly through improved access for formal credit (your local bank manager may look more kindly on your request for a loan if she is aware that most of your income is a stable cheque guaranteed by the government). Even if farms are not credit-constrained, subsidies may represent a cheaper source of financing than the credit available from your local bank.
Subsidies can also positively affect farm behaviour under uncertainty through a wealth effect. Farmers may be more willing to expand production through certain types of activities that would otherwise be viewed as too risky in the absence of the guaranteed income from the direct payment.
On the other hand, subsidies may negatively affect farm productivity because they distort the production structure of recipient farms. An obvious example is a coupled subsidy which keeps farmers engaged in a loss-making enterprise (which might be keeping suckler cows, for example) simply in order to draw down the subsidy.
Subsidies may give rise to technical inefficiency if higher profits lead to slack, a lack of effort and disinclination to seek cost-reducing methods. Subsidies also lead to a soft budget constraint, meaning that farmers might be inclined to over-invest leading to inefficient use of resources. The number of shiny new tractors on Irish farms despite the evidence of low incomes testifies to this effect. More generally, subsidies help to keep existing resources in the industry and slow down the rate at which resources are reallocated to more productive uses in response to new technologies or market conditions.
Thus, whether the positive or negative effects dominate is an empirical question. By chance, a new study by three agricultural economists, Marian Rizov, Jan Pokrivcak and Pavel Ciaian, attempts to answer exactly this question. The authors investigate the impact of CAP direct payments on farm productivity in the EU-15 member states (the absence of sufficiently long data time series precluded covering the new member states).
Measuring the productivity effects of direct payments
Previous attempts to examine the relationship between CAP subsidies and farm productivity used a two-stage approach. First, a production function is fitted to farm-level data and estimates of productivity are derived. In a second stage, those productivity estimates are regressed on subsidies to try to identify the impact of the subsidies on the level of productivity.
The problem with this approach is that, if it is assumed that subsidies affect productivity, then subsidies should also be included in the first stage estimation of productivity levels, as otherwise the estimates will be biased.
The authors of the new study use a methodology which explicitly incorporates subsidies in estimating farm-level productivity. They also controlled for the selection bias which might arise due to the exit of farms over time (it is likely that those farms that exit have lower productivity on average than those farms that remain, so failing to take this effect into account will also bias the productivity estimates). Their methodology further takes account of the well-known simultaneous relationship between productivity levels and input demands (the choice of inputs will be correlated with the farm’s productivity level). With their approach, they are able to test for the impact of direct payments both before and after decoupling was introduced in the period 2005-06.
For their study the authors use FADN data for six main farm types for the 15 old member states over the period 1990-2008 (for Austria, Finland and Sweden which entered the EU in 1995, the period of analysis is 1996-2008). Their empirical strategy is to run regressions within the six farm-type samples for each country, which gives them 83 farm-type country samples; this approach allows for flexibility and variation in technology choices across farm systems and countries.
In order to test for the relationship between farm productivity and subsidies, the authors first estimate farm-level productivity levels and growth rates. Productivity is measured as the growth in output after all measured inputs are accounted for (a measure known as total factor productivity, or TFP). These are aggregated to national levels using output weights, thus giving bigger weight to productivity levels/growth rates on larger farms.
The calculated average annual percentage TFP growth rates by member state are shown in the figure. The southern European countries Italy, Spain and Portugal all show rapid TFP growth over the period; perhaps surprisingly, the countries in north-west Europe (Belgium, Netherlands and Ireland) as well as the Nordic countries (Finland, Sweden and Denmark) show negative productivity growth (either 1990-2008 or 1996-2008 depending on the date of EU accession).

How important are the productivity effects of direct payments?
The main purpose of the study was to identify the impact of CAP direct payments on productivity levels and growth. Here the authors find evidence that coupled payments (prior to 2005) had a clear negative effect on productivity (the finding is statistically significant even if economically the magnitude of the effect is not great – a doubling of subsidies leads to a reduction of between zero and 3.7 per cent in TFP depending on the country).
However, for the period when subsidies were decoupled, a more varied pattern of results is found. For ten of the EU-15 countries there is a positive relationship between subsidies and productivity although this relationship is statistically significant for only six countries for both productivity level and growth. Overall, they conclude that decoupled subsidies after 2005 either have no effect or a small positive effect on productivity in the majority of EU-15 countries.
These findings are consistent with the study by Kazukauskas, Newman and Sauer of Danish, Dutch and Irish farms using a similar methodology but with a uniform production function for each of the member states. They also found that decoupling had a positive and significant effect on productivity.
These are net effects; the methodology does not distinguish between the separate effects of the allocative and technical inefficiency losses and the investment-induced productivity gains. What the results suggest is that, with decoupling, the allocative and technical inefficiency losses are reduced, and/or the positive investment effects due to the interaction of the subsidy with market imperfections are increased. However, in all cases, the economic importance of the effects identified is very small.
Going back to my Twitter debate with Martin Crowe, the results of this study suggest that, in the era of decoupled payments, Martin’s point that the “70% [of income coming from direct payments] gives the security and confidence to try and improve the 30%” trumps my concern that “If 70% of your income is coming as a cheque in post there is less incentive to innovate to grow the remaining 30%”. However, the economic importance of the productivity effect is small.
Whether decoupled subsidies impact on output levels by encouraging higher levels of investment and/or variable input use is also of interest but this is not a question addressed by this study.

Posted on

More thoughts on the European Innovation Partnership for Agriculture

One of the more widely-welcomed elements of the Commission’s legislative proposals for the CAP post 2013 was a greater emphasis on promoting innovation. In an earlier post in January, I discussed the three main components of this strategic emphasis, namely:
• Continued Pillar 2 support for investment in physical assets and farm extension services.
• A new European Innovation Partnership instrument for agricultural productivity and sustainability (EIP-A) also in the Rural Development Pillar.
• Increased funding amounting to €4.5 billion for agricultural and food research under the Commission’s Horizon 2020 research programme for research and innovation on food security, the bio-economy, and sustainable agriculture.
Since that previous post two developments of note have taken place. The Commission published its Communication on EIP-A in February. And this week science ministers gathered in the Competitiveness Council, after long and hard negotiations, gave their backing to a general structure for the Horizon 2020 research programme.
In EU jargon they agreed a ‘partial general approach’ (no kidding!) which is an agreement on the essential elements of a legal act, pending the opinion of the European Parliament and the relevant discussions on the EU’s multi-annual budget. One of the main sticking points was the demand of the new member states for ‘equal access’ and ‘fair conditions for newcomers’ that would put an end to the ‘closed clubs’, by which they mean the dominance of western EU states in winning the bulk of funding.
European Innovation Partnership – Agriculture
The Commission Communication in February presented the conception of the EIP-A but added little to the limited information in the public domain as to exactly how this Innovation Partnership is supposed to work.
The objectives and rationale for the Partnership are repeated. The growing demand for food will put further stress on the environment and natural resource use. Therefore a shift towards a different growth path is needed in order to establish a competitive and sustainable production of food, feed, fibre, biomass and biomaterial. At the same time, the increase in output must go hand in hand with improved economic viability for primary producers who have suffered a declining share of value-added in the food chain over the past decade. Without greater farm profitability, ecological sustainability will become even more challenging.
The existence of a gap between the provision of research results and the application of innovative approaches to farming practice as a justification for the EIP is again underlined. New approaches take too long to arrive on the ground, and the needs of practical farming are not communicated sufficiently to the scientific community. Thus, important innovations are not implemented on the necessary scale, and relevant research fields do not always receive the attention they require.
Targets and priority areas for research
The main novel element in the Communication was the identification of two headline targets for the EIP:
• As an indicator for promoting productivity and efficiency of the agricultural sector, the EIP aims to reverse the recent trend of diminishing productivity gains by 2020.
• As an indicator for the sustainability of agriculture, the EIP aims to secure soil functionality in Europe at a satisfactory level by 2020. Soil functionality encompasses the productive capacity of soils and its key roles in climate change mitigation and adaptation and eco-system stability.
A number of indicative priority areas for research and innovation based on the stakeholder consultation are set out, while making clear that the list should not pre-empt the content of innovation actions on the ground. The areas include:
• Increased agricultural productivity, output, and resource efficiency
• Innovation in support of the bio-based economy
• Biodiversity, ecosystem services, and soil functionality
• Innovative products and services for the integrated supply chain
• Food quality, food safety and healthy lifestyles
These headings broadly overlap with the objectives set out in the proposed Council Decision establishing the specific programme implementing Horizon 2020 published towards the end of last year. The headline targets proposed for the EIP arguably capture only a sub-set of these proposed activities.
The next step in the establishment of the EIP-A is the preparation of a Strategic Implementation Plan; no details are provided as to how this will be done. The main concrete proposal in the Communication is that an EIP network facility will be established under the umbrella of the Rural Development Network. This network should animate activities at Union, national, regional, and local level. It will encourage the establishment of operational groups and inform about the opportunities provided by Union policies.
The Communication notes that “The timely establishment of an EIP network is needed to ensure early information of actors and stakeholders concerning opportunities for innovative action.” However, although the Communication was launched in February, a search of the ERDN website when writing this post today revealed no further information about the EIP network, how it will be initiated or who can join it.
Productivity change

The productivity target for the EIP-A will be measured as ‘total factor productivity’. This measure has solid foundations, but the data requirements to monitor it are demanding and currently only available for a limited number of EU-15 countries. Work on calculating TFP growth is undertaken by the Groningen Growth and Development Project supported by an EU FP7 grant, so one assumes that this will be the data source for the EIP productivity indicator.
The latest information on their website covers TFP growth in the period 1980 through 2007 for the 10 EU-15 countries for which growth accounting could be performed, namely: AUT, BEL, DNK, ESP, FIN, FRA, GER, ITA, NLD & UK. The graph shows TFP growth trends in agriculture as well as the food industry and all industries as calculated by this project (the trends are normalised to base 1995 = 100). It confirms the slow-down in agricultural productivity that the EIP-A sets out to reverse.

The rapid technological development of EU agriculture since 1980 compared to other industries is striking. The slow-down since 1999 could be due to a number of reasons. Reduced investment in agricultural research during the period of structural surpluses in the 1980s and 1990s might be one explanation. The slowdown could also be due to the increasing stringency of environmental, food safety, animal welfare and health regulations with which the sector must comply. The TFP trend for the EU food industry is even more alarming, suggesting that TFP is actually lower today compared to the early 1990s.
There is no doubt that a gap exists between laboratory findings and farm-level uptake. However, whether this gap is due to structural barriers between research and practice, or due to the wide distribution of management abilities among farmers, seems to me unproven. It is hard to disagree with the principles on which the EIP-A is based, but whether it will play an important role in reversing these TFP trends remains unclear.

Photo credit Michael Trolove