The UK’s labor productivity growth slowdown
An analysis of the labor productivity growth slowdown in the United Kingdom
since the start of the 2008 financial crisis
Martijn Kunst 6052460
2
Table of Contents
1. Introduction1.1. Facts and figures 3
1.2. Labor productivity 7
2. Literature and Theory
2.1 Drivers 9
2.2 Labor hoarding 10
2.3 Impaired reallocation of capital 11
2.4 Lower capital-labor ratio and labor market flexibility 11
2.5 Intangible investments 13
2.6 Shift in the structure of the economy 14
2.7 Temporary versus permanent decline 17
3 Channels
3.1 Intangible investments 18
3.2 Labor hoarding 18
3.3 Labor market flexibility 18
3.4 Capital-labor Ratio 19
3.5 Impaired reallocation of capital 19
4 Methodology 4.1 Regression 21 4.2 Data 22 5 Results 25 6 Summary 32 7 Literature list 33
3
1. Introduction
1.1 Facts and Figures
Labor productivity is one of the most important macroeconomic variables in modern economics. It is one of the key drivers of long-run economic growth; improvements in labor productivity help firms to produce at lower costs and therefor making them more competitive. These lower costs also translate into lower prices, which then increases households’ standard of living. However, in the UK, labor productivity levels took a bit hit during the crisis, decreasing by almost 4% from its pre-crisis peak. In 2013 (5 years after the initial downturn) the labor productivity level in the UK still hovered around the same level as just after the crisis, lagging an estimated 16% below the implied level with the pre-crisis growth rate.
Graph 1, labor productivity per hour in the UK
From the above graph, one can see that GDP per hour has been steadily increasing in the UK for the past 55 years. However, since the 2008 crisis, labor productivity levels have remained stable; showing zero or even negative labor productivity growth. This is a clear break in the trend shown since 1950. With labor productivity consisting of employment and output, it is important to see what these two factors have done since the 2008 crisis.
10,00 15,00 20,00 25,00 30,00 35,00 40,00 45,00 50,00 55,00 19 50 19 55 19 60 19 65 19 70 19 75 19 80 19 85 19 90 19 95 20 00 20 05 20 10
GDP per Hour in the UK
4
In the first quarter of 2014, total number of people employed in the UK increased by 345,000 to 30.54 million. The rate of unemployment fell from 7.8 to 6.6 percent, meaning a decline of 1.2 percentage points from a year earlier. From these numbers one could conclude that the UK economy is recovering from the recession and performs very well. The increase in employment, however, is not matched by the UK’s production, as shown in graph 2. GDP in 2013 was still below its peak in 2008 and has shown a decline compared to 2012, showing that the recovery is still very weak.
Graph 2, UK GDP in € millions
Employment developments in the UK are shown in graph 3, with 2008 the only year in which employment was declining. An increase in employment is not automatically a bad development, but the point in this case is that these extra workers are not translated in a higher production. In other words: there are more people producing fewer goods, and that really is a negative development. 0,00 500.000,00 1.000.000,00 1.500.000,00 2.000.000,00 2.500.000,00 19 71 19 74 19 77 19 80 19 83 19 86 19 89 19 92 19 95 19 98 20 01 20 04 20 07 20 10 20 13
GDP of UK in €millions
GDP in €millions5 Graph 3, Employment level in the UK, x1000’s
Data from the Office of National Statistics (ONS) also suggests that labor productivity took a much bigger hit compared to past recessions in the UK, as shown in graph 4. Which also suggests that this crisis is different compared to previous ones.
Graph 4, development of output per worker after a recession (number of quarters on the horizontal axis) 22.000 23.000 24.000 25.000 26.000 27.000 28.000 29.000 30.000 19 50 19 53 19 56 19 59 19 62 19 65 19 68 19 71 19 74 19 77 19 80 19 83 19 86 19 89 19 92 19 95 19 98 20 01 20 04 20 07 20 10 20 13
Employment Levels in the UK
Employment Levels 90 95 100 105 110 115 120 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18
Output per worker
Quarters after start of recession
1973Q2 1979Q4 1990Q2 2008Q1
6
If labor productivity growth had declined in other countries as well, the problem would be less severe and might be contributed to some worldwide economic slowdown. However, it seems as if the UK is the only country experiencing a slowdown in labor productivity growth.
Graph 5, Labor productivity across different countries
Graph 5 compares the UK labor productivity to the other 4 major euro area economies: France, Germany, Italy and Spain. France and Germany have both shown a slight decrease between 2007 and 2009 but are now recovering. Italy has shown zero labor productivity growth since the year 2000 and this did not change much during the crisis. Spain is also showing an upward trend, making the UK the only country that shows a decline in labor productivity. Graph 6 shows UK’s labor productivity, compared to the US and Japan. Again, the UK is the only country with a declining labor productivity growth, as both the US and Japan show an upward trend.
30 35 40 45 50 55 60
GDP per hour, across Euro area countries
7 Graph 6, Labor productivity across different countries
So the UK is the only country showing negative labor productivity growth since 2008, is this a problem?
1.2 Labor Productivity
It is normal for labor productivity to decline during recessions; as there is less demand for
products, hence output decreases and labor productivity too (following 𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃 =𝑂𝑂𝑂𝑂𝑂𝑂𝑂𝑂𝑂𝑂𝑂𝑂
𝐼𝐼𝐼𝐼𝑂𝑂𝑂𝑂𝑂𝑂 ).
After some time, firms shed workers or, if the economy picks up again; demand and output recover, raising labor productivity. However, ever since the UK economy got hit by the
2007/2008 financial crisis, labor productivity has not yet recovered, instead it still hovers around its 2009 level and has not shown much improvement since then.
It is important to find out why labor productivity growth has not increased because it is the key driver of long-run economic growth. As improvements in labor productivity help firms to produce at lower costs and therefor makes them more competitive. These lower costs also translate into lower prices, which increases households’ standard of living. In addition the productivity puzzle is also interesting from a monetary policy point of view, as the Bank of England usually raises the interest rate when employment is high, because when more people are working and receive a wage, consumer spending will increase, which in turn increases prices. An
25 30 35 40 45 50 55 60 65 70 19 90 19 91 19 92 19 93 19 94 19 95 19 96 19 97 19 98 19 99 20 00 20 01 20 02 20 03 20 04 20 05 20 06 20 07 20 08 20 09 20 10 20 11 20 12
GDP per hour in Japan, UK and US
8
increase in the interest rate will counteract these increased prices by making money more expensive (for example, people will get a higher interest rate on their savings account; making them save more and spend less) and reducing spending. This prevents the economy from overheating. However in the current situation an increase in the interest rate because of the high employment could kill a recovery of the economy.
Some explanations for this puzzle have already been proposed, one of them is mismeasurement of GDP; this view argues that the productivity puzzle is caused by deficiencies in the measurement of GDP and/or labor market statistics. However the ONS argues that this “Would require revisions in these statistics on a scale which was both implausible and for which we have to date been able to find no evidence“ (Grice, 2012). Luckily, there are more explanations for the behavior of labor productivity and this paper will try to find out which ones are credible explanations and which ones are not. In addition, it will make a comparison between the UK and other countries to see where the UK deviates in its development and may therefor cause the difference observed between the UK and other OECD countries.
After this introduction, section 2 will focus on the theories that explain labor productivity growth behavior and possible causes for the slowdown in the UK. Section 3 will explain the channels by which the various proposed theories actually influence labor productivity growth. Section 4 explains the data and methodology used in the analysis after which section 5 shows and interprets the results.
9
2. Literature and Theory
2.1 Drivers of Labor Productivity
The UK’s productivity puzzle is not the first of its kind, in the 1970’s the US and other countries also suffered from a slowdown in labor productivity growth. Possible and viable explanations put forward by researchers for this were changing demographics (by that time the baby boom generation was fully absorbed into the workforce), deteriorating levels of education, lower investment (so lower capital-labor ratio) and also a lower public capital growth rate (Munnell, 1990). This last point is also made by Aschauer (1989), who finds that public capital has “an important role for the net public capital stock in the ‘productivity slowdown’ of the last fifteen years”. Lynde and Richmond (1993) also conclude that as much as 40 percent of the productivity decline in the 1970’s was caused by a lower public capital-labor ratio. This is however contradicted by Holtz-Eakin (1992), who does not find a significant quantitative connection between public capital and productivity after controlling for unobserved state-specific effects.
In addition, Sondermann (2012) finds that increased regulations have a negative effect on labor productivity growth and that increases in R&D investment and ICT capital have positive effect on, especially, the manufacturing sector, but not on the service sector (which involves many lower skilled jobs that benefit less from highly skilled labor).
New Zealand too is suffering from a low labor productivity growth. Research by de Serres, Yahiro and Boulhol (2014) concluded, however, that capital and human investment are on par with similar (advanced) countries and that the country has very supportive policies for education, tax and regulations. However, a low level of knowledge-based capital and poor geographical location and connections may be causing the productivity puzzle in New Zealand.
In Timmer et al (2010) labor productivity is analyzed using the KLEMS growth accounting database (KLEMS is an abbreviation of the inputs used: capital (K), labor (L), energy (E), material inputs (M) and service inputs (S)). The focus is in particular on the difference between the European Union and the United States. They find that Europe’s slower growth is mainly due to a productivity slowdown in the manufacturing sector and the failure of fully benefitting from new ICT developments.
In addition they find that labor productivity has had three important drivers during the last thirty years: direct investments in information and communication technology, changes in labor composition (due to greater demand for higher skilled workers) and multifactor productivity growth.
10
The ONS also identifies certain factors that are important for determining underlying productivity growth; these five drivers of productivity are identified in their productivity handbook as:
• Investment, especially in capital
• Innovation, which can translate into new technology • Skills, more skilled employees can produce more
• Enterprise, seizing of new opportunities by existing firms and start-ups
• Competition increases incentives for firms to become more efficient and productive.
(
http://www.ons.gov.uk/ons/guide-method/method-quality/specific/economy/productivity-measures/productivity-handbook/index.html)
2.2 Labor hoarding
This view was put forward by Martin and Rowthorn (2012), they argue that as demand lowers, firms have excess capacity and are able to run with fewer workers. Firms retain labor because they usually possess firm-specific skills, which are costly for the firms to replace after they have been fired when demand recovers.
In addition, one might assume that a firm uses low-productivity variable labor to produce output and highly productive and skilled labor to manage the firm (overhead). As production decreases it will fire the variable labor and hold on to the higher skilled labor (which is more expensive to hire, once fired). Once recovery picks up, production will increase and firms will hire additional low-productivity variable labor, thus explaining why this recovery has a low labor productivity growth rate (Broadbent, 2012).
This argument can thus be split in two parts; the first paragraph takes place at the start of the recession, while the second paragraph is more applicable to the recovery phase in a recession.
This labor hoarding view is deemed unlikely by a number of papers, as they argue that the economy has been in a recession for almost five years now and firms do not hoard labor for that long. Furthermore, the increase in employment is mainly due to job creation. This raises the following question; if firms are labor hoarding, why hire new people? (See graph 7).
11
Graph 7, Private sector job creation (graph taken from the Office of National Statistics (ONS)) 2.3 Impaired reallocation of capital (also known as forbearance by banks)
There is also the possibility that the reallocation of capital is not working optimal, due to the behavior of banks. As banks are reluctant to write off loans to more established low-productivity firms and do not like to lend capital to start-ups with higher productivity, which they perceive as being more risky. A given aggregate quantity of capital may be allocated in different ways across firms of heterogeneous efficiency, but allocating too much capital to inefficient firms for example will diminish aggregate productivity. In short, capital stays at low productivity firms and is not moved to more productive firms, lowering aggregate labor productivity. Field and Franklin (2013) find evidence that the level of productivity below which firms exit an industry has fallen over the recession.
There is not much research done on this topic yet, however it seems likely that when banks are under pressure to restructure their balance sheet, they are reluctant to write off loans, as this would further damage their balance sheet.
2.4 Lower capital-labor ratio and labor market flexibility
Both the lower capital-labor ratio and the labor market flexibility explanations are caused by lower real wages, they, however, use different transmission mechanisms:
1. Lower labor costs and higher cost of capital (due to credit market tightness) makes it more attractive for firms to shift resources away from capital and into labor, therefore lowering the capital-labor ratio. This was not an issue with previous recessions; however, reforms to union strength and welfare have made wages more sensitive to negative demand shocks. Van Reenen and Pessoa (2013) find that this actually makes quite a difference, explaining around 60% of the change in labor productivity.
12
2. Investment has gone down during the crisis; therefore the current capital stock is smaller. If workers have less capital at their disposal they are usually less productive.
3. Lower labor costs also make it more attractive for firms to retain or hire lower productive labor. These lower wages are not only due to individuals getting fired and having to accept a new job with lower pay, but also by wage reductions within jobs themselves (Blundell et al, 2013).
So the question is why do firms not substitute towards capital and increase their labor
productivity? The answer may lie in the availability of capital; in graph 8, the total amount of outstanding debt is plotted. One can see that after the crisis, this amount has been steadily decreasing.
Graph 8, Net debt of all corporate sectors in the UK
In addition to this, the interest rate SMEs pay on their debt has been steadily increasing since 2009 (see graph 9). Especially smaller SMEs (those with a annual debit account turnover on the main business account of less than £1 million) have seen their average interest rate increase from 3.75% in July 2009 to 4.75% in February 2012. This is an important fact, as SMEs are
responsible for 59.3% of private sector employment according to the “Federation of Small Businesses”. 1000000 1050000 1100000 1150000 1200000 1250000 1300000 1350000 1400000 1450000 1-ju n-09 1-se p-09 1-de c-09 1-m rt -10 1-ju n-10 1-se p-10 1-de c-10 1-m rt -11 1-ju n-11 1-se p-11 1-de c-11 1-m rt -12 1-ju n-12 1-se p-12 1-de c-12 1-m rt -13 1-ju n-13 1-se p-13 1-de c-13 1-m rt -14
Net debt of all Corporate Sectors in UK
13 Graph 9, interest rates paid by SME, ordered by size
This fits nicely in the low investment and low capital-labor-ratio story. Even though labor is getting more expensive, firms struggle to acquire finance and are not able to invest in new capital. The consequence is that labor productivity does not increase, while labor costs do.
2.5 Intangible investments
Goodridge, Haskel and Wallis (2013) find that intangible investments can explain a significant portion of the UK’s productivity gap, which they define as the difference between the actual labor productivity level and the implied level by the pre-crisis trend line. They argue that when spending on intangibles forms a considerable contribution to investment, actual investment may be higher, reducing the gap between the implied labor productivity level and the actual value, thereby reducing the productivity gap. They also find evidence of this in earlier studies (Haskel et al, 2009). Or as Pessoa and van Reenen (2013) put it: when there is a high growth in intangible investments, productivity is usually underestimated (because of intangible investments being excluded from GDP calculations and thus the real value of GDP is higher than the one calculated) and vice versa: a low growth in intangible investment gives lower productivity. In this paper they find evidence for an increase in intangible investment throughout the crisis, implying an underestimation of the actual level of labor productivity growth.
3,00 3,50 4,00 4,50 5,00 5,50
14
This argument can be considered as a “measurement error”, however here it is not so much an error, as it has more to do with the methodology of measuring labor productivity.
2.6 Shift in the “structure” of the economy
The UK saw a strong decline in the share of manufacturing firms in GDP between 1995 and 2008, after which it suddenly remained constant until 2012, as can be seen in graph 10:
Graph 10, share of manufacturing firms in GDP (%) in the UK
Graph 11, share of manufacturing firms in GDP (%) in Spain and New Zealand
This pattern is similar in countries like Spain and New Zealand (graph 11), both of which also suffer from weak labor productivity growth. In addition also Australia and Luxembourg show a
0 5 10 15 20 19 90 19 91 19 92 19 93 19 94 19 95 19 96 19 97 19 98 19 99 20 00 20 01 20 02 20 03 20 04 20 05 20 06 20 07 20 08 20 09 20 10 20 11 20 12
Share of Manufacturing in the UK
Share of Manufacturing 0 5 10 15 20 19 90 19 91 19 92 19 93 19 94 19 95 19 96 19 97 19 98 19 99 20 00 20 01 20 02 20 03 20 04 20 05 20 06 20 07 20 08 20 09 20 10 20 11 20 12
Share of Manufacturing in SPA & NZ
15
slightly declining line. For other countries it looks very different, for example France, Germany and Italy, as shown below (graph 12):
Graph 12, share of manufacturing in GDP (%) in France, Germany and Italy
These countries did not see the large decline and there are similar results in other countries, they all show increasing or constant shares of manufacturing in GDP, but no decline. One can see that since 2009, the number of defaults has shown a roughly downward trend. In graph 14, the number of defaults is plotted.
This could mean that the shift from manufacturing to services was stopped due to the fact that less manufacturing firms went into bankruptcy, which could mean that more firms with low productivity were allowed to stay alive. The following graph (graph 13) shows that up until 2009, service sector firms where more productive then manufacturing sector firms, this means that the increase in aggregate labor productivity was, at least partially, caused by the increasing share of service sector firms relative to manufacturing sector firms. However, already before the crisis it can be seen that productivity between the two sectors is converging; therefor this
particular factor in the productivity slowdown might have also happened without the 2008 financial crisis.
This makes that a change in the share of manufacturing seems the most likely explanation for the slowdown in labor productivity growth in the UK compared to other countries.
0 5 10 15 20 25 30 19 90 19 91 19 92 19 93 19 94 19 95 19 96 19 97 19 98 19 99 20 00 20 01 20 02 20 03 20 04 20 05 20 06 20 07 20 08 20 09 20 10 20 11 20 12
Share of Manufacturing in GDP
France Germany Italy16 Graph 13, Productivity per sector in the UK, GDP per hour
Graph 14, number of defaults in the UK
18 20 22 24 26 28 30
Productivity by Sector
Services Manufacturing 20,00 30,00 40,00 50,00 60,00 70,00 80,00 90,00 100,00 2004 Q1 2004Q3 2005Q1 2005Q3 2006Q1 2006Q3 2007Q1 2007Q3 2008Q1 2008Q3 2009Q1 2009Q3 2010Q1 2010Q3 2011Q1 2011Q3 2012Q1 2012Q3 2013Q1 2013Q3Number of Defaults in the UK, index numbers
17 2.7 Temporary versus permanent decline
Generally speaking, there are two sides in the debate on the productivity puzzle: those who think that the decline is temporary and those who think that it is permanent, also known as supply-side-pessimists. The latter view argues that the decline in labor productivity is permanent and structural, either by being linked to the financial crisis or by mismeasurement of GDP and that potential output is already close to actual output. In this case quantitative easing will only increase inflation and not have any stimulating effect on the economy (King, 2013)(Broadbent, 2012).
Possible causes for a decreased economic capacity, either permanent or temporary, might be an impaired ability of firms to anticipate recoveries in demand due to cuts in investment, reduction of R&D spending due to uncertainty, loss of human capital (as people are jobless for long periods of time) or lower labor quality as firms cut back on funding for training (Patterson, 2012).
A study by the Bank of England argues that this banking crisis reduces average long-run level of capital per worker by about 1% per year of crisis and reduces long-run growth rate of labor productivity by between 0.84% and 1.1% (Oulton and Sebastia-Barriel, 2013)
This is rejected by Pessoa and van Reenen (2013) who argue that:
1. The UK is not the only country suffering from a fall in labor productivity; other EU countries suffered the same fate. US productivity did a much better job, however this is mainly because of much higher unemployment and not due to increased GDP.
2. The UK has experienced a large increase in labor productivity since 1979, outpacing countries like Germany and France. Aghion et al (2013) argue that, at least part of these productivity improvements are due to policy reforms, like increased product market competition, labor market flexibility and more powerful independent regulating institutions. If these reforms gave rise to a steady and positive growth of the labor productivity, it seems unlikely that they have suddenly disappeared after the crisis.
However there are also a large number of people who think that this labor productivity slowdown is temporary and caused by demand side problems, a number of possible explanations using this idea have been brought forward and these are the explanations this paper will be focused on.
18
3. Channels
3.1 Intangible investments
The aforementioned explanations work through a number of channels. The first one is intangible investment growth rates. The effect of intangible investments (such as R&D) is usually not immediate; there is a lag between the investment and the benefits (higher labor productivity for example) that it brings. This means that a decline in the intangible investment growth means lower intangible investment levels and after some years to lower productivity growth rates compared to what it could have been without the decline in intangible investment growth rate.
3.2 Labor hoarding
When assessing the labor hoarding explanation, the goal is to see if firms are holding on to skilled labor, even though this means that they have excess capacity. This means we have to find a way to distinguish between workers with different skill sets. A way to adjust labor quantities for a heterogeneous workforce is through skill composition. This measure controls for various differences between workers (age and gender for example), but is mainly driven by differences in skill levels (Haskel et al, 2013). Therefor an increasing skill composition index means that the average skill level per employee is increasing. Also, lower labor costs make it more attractive to retain employees, as they are relatively inexpensive.
3.3 Labor market flexibility
It is almost impossible to list all the factors that influence labor market flexibility, however, how difficult it is to hire and fire workers may be a good indication of how flexible a certain labor market is. A good proxy to look at is there for employment protection. High labor protection means it is difficult for firms to fire workers and therefor the quantity of labor will react slower to changes in demand. This can be measured by analyzing the cost and procedures necessary for firing fixed-contracts workers or hiring them. The OECD employment protection index, for example, summarizes all employment protection legislation into a single index
(http://www.oecd.org/employment/emp/oecdindicatorsofemploymentprotection.htm). When a
country’s index level is high, it means that it is more difficult to fire workers and the country probably has a lower adjustment rate of the employment level to changes in demand, while a low level of the index means the opposite. Also nominal compensation per employee should capture some of this effect as lower wage costs mean that employers are more willing to retain lower productivity labor.
19 3.4 Capital-labor ratio
The lower capital-labor ratio can be caused by a number of factors, first of all: the uncertainty of the 2008 financial crisis caused a decrease in investment growth, this may lead to lower capital stock levels in the present and, hence a lower capital-labor ratio. In addition, because this 2008 crisis was also a financial crisis, it has increased the cost of capital, as it is more difficult for firms to acquire funding, as banks are reluctant to lend money, making it more attractive to substitute away from the relatively expensive capital towards relatively cheaper labor. As a proxy for these credit constraint conditions, domestic credit by financial sectors is used. The larger the amount of domestic credit provided by the financial sector, the easier it should be for firms to obtain credit (due to increased supply).
Also, since the crisis, labor costs have remained relatively stable and are still beneath the 2007 level (see graph 15). This gives another incentive to firms to substitute towards labor and away from capital.
Graph 15, Real compensation per employee, deflated by GDP
All of these effects should be captured by growth rate of capital deepening, which measures the growth in the amount of capital per worker.
3.5 Impaired reallocation of capital
Banks have been reluctant to lend money because they are repairing their balance sheets; this makes it difficult for small and medium size companies (SMEs) to acquire the capital needed to grow. As these SMEs are usually more productive than larger established companies, this keeps
0 20 40 60 80 100 120 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012
20
aggregate productivity levels below what otherwise would be the case. Research has shown that ICT is an important driver of increases in productivity (van Ark et al, 2003), ICT capital deepening is there for used to asses developments in the amount of ICT capital per worker.
21
4. Methodology
4.1 Regression
This paper will use a panel data regression to determine the factors that have an effect on labor productivity in OECD countries. As it seems unlikely that the effects of intangible investments are instant, a five-year lag is added to make it more realistic. The data used will be from 1990 till 2013 and the regression will have the following form:
(𝐿𝐿𝐿𝐿𝐿𝐿𝑃𝑃𝑃𝑃 𝑝𝑝𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃 𝑔𝑔𝑃𝑃𝑃𝑃𝑔𝑔𝑃𝑃ℎ)𝑂𝑂,𝑖𝑖 = 𝛼𝛼𝑖𝑖 + 𝛽𝛽1(∆𝑅𝑅𝐿𝐿𝑅𝑅)𝑂𝑂,𝑖𝑖+ 𝛽𝛽2(∆𝑅𝑅𝐿𝐿𝑝𝑝𝑃𝑃𝑃𝑃𝐿𝐿𝐶𝐶 𝑃𝑃𝑑𝑑𝑑𝑑𝑝𝑝𝑑𝑑𝑑𝑑𝑃𝑃𝑑𝑑𝑔𝑔)𝑂𝑂,𝑖𝑖 + 𝛽𝛽3(∆ 𝐼𝐼𝑑𝑑𝑃𝑃𝐿𝐿𝑑𝑑𝑔𝑔𝑃𝑃𝐿𝐿𝐶𝐶𝑑𝑑 𝑃𝑃𝑑𝑑𝑃𝑃𝑑𝑑𝑖𝑖𝑃𝑃𝑖𝑖𝑑𝑑𝑑𝑑𝑃𝑃)𝑂𝑂−3,𝑖𝑖+ 𝛽𝛽4(∆𝐻𝐻𝑅𝑅𝐼𝐼)𝑂𝑂+ 𝛽𝛽5(𝐸𝐸𝑃𝑃𝐼𝐼)𝑂𝑂,𝑖𝑖 + 𝛽𝛽6(∆𝑆𝑆ℎ𝐿𝐿𝑃𝑃𝑑𝑑 𝑃𝑃𝑜𝑜 𝑖𝑖𝐿𝐿𝑑𝑑𝑃𝑃𝑜𝑜𝐿𝐿𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑑𝑑𝑔𝑔 𝑃𝑃𝑑𝑑 𝐺𝐺𝐺𝐺𝑃𝑃)𝑂𝑂,𝑖𝑖 + 𝛽𝛽7(∆𝐼𝐼𝑅𝑅𝐼𝐼 𝑃𝑃𝐿𝐿𝑝𝑝𝑃𝑃𝑃𝑃𝐿𝐿𝐶𝐶 𝑃𝑃𝑑𝑑𝑑𝑑𝑝𝑝𝑑𝑑𝑑𝑑𝑃𝑃𝑑𝑑𝑔𝑔)𝑂𝑂,𝑖𝑖+ 𝛽𝛽8(𝑂𝑂𝑝𝑝𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑖𝑖𝑖𝑖)𝑂𝑂,𝑖𝑖 + 𝛽𝛽9(∆𝐹𝐹𝑃𝑃𝑑𝑑𝐿𝐿𝑑𝑑𝑃𝑃𝑃𝑃𝐿𝐿𝐶𝐶 𝑖𝑖𝑑𝑑𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃 𝑖𝑖𝑃𝑃𝑠𝑠𝑑𝑑) + 𝜀𝜀𝑂𝑂,𝑖𝑖 With: Labor Productivity Growth
Labor productivity growth as the growth in GDP per hour worked. Data is from the Conference Board’s “Total Economy Database” (TED).
RLC Real labor costs per Employee, data retrieved from the Eurostat Annual
Macro-Economic Database (AMECO).
Capital deepening Growth rate of capital deepening, data from the Conference Board’s TED. Intangible
Investment Growth
Intangible investment growth, data from the Intan-invest database on intangible investment.
HCI Human Capital Index, data available in the Penn World Table, based on
years of schooling and returns to education.
EPI Employment Protection Index measures the amount of protection workers
have against getting fired. Share of
Manufacturing in GDP
The value added by the manufacturing sector divided by total value added by the whole economy.
ICT capital deepening
Growth in the amount of capital per worker.
Openness of the economy
The sum of the absolute values of the share of exports in GDP and share of imports in GDP.
Financial Sector Size
Domestic credit provided by financial sector (% of GDP), gathered from the World Bank.
22
The database includes all 33 OECD countries,
Australia Austria Belgium Canada Chile Czech Republic Denmark Estonia Finland France Germany Greece Hungary Iceland Ireland Italy Japan Luxembourg Mexico Netherlands New Zealand Norway Poland Portugal Slovakia Slovenia South Korea Spain Sweden Switzerland Turkey United Kingdom United States
The resulting coefficients will show which factors have a significant effect on labor productivity growth in advanced economies. These results will then be compared to developments in the UK and the proposed explanations of the labor productivity puzzle there.
Due to the fact that not all of the required data is available in every country, the countries that will be used in this first regression are: Austria, Belgium, Czech Republic, Denmark, Finland, France, Germany, Ireland, Italy, Netherlands, Slovenia, Spain, Sweden, UK and the US. This is due to the scarcity of data on intangible investments, which is only available for the years 1995-2010.
4.2 Data
Labor productivity growth is calculated by taking GDP per hour worked, this also controls for the increasing amount of part-time workers that pollutes productivity per worker. Growth rates are calculated with a simple formula:
100% ∗(𝑑𝑑𝑑𝑑𝑔𝑔 − 𝑃𝑃𝐶𝐶𝑃𝑃)𝑃𝑃𝐶𝐶𝑃𝑃 = 𝑔𝑔𝑃𝑃𝑃𝑃𝑔𝑔𝑃𝑃ℎ 𝑃𝑃𝑑𝑑 %
Real Labor Costs are taken from the AMECO database. Growth rates are, again, calculated with the formula shown before.
Openness of the economy is calculated by summing the percentages of exports and imports in GDP in the Penn World Table using the following formula:
(% 𝑑𝑑𝑒𝑒𝑝𝑝𝑃𝑃𝑃𝑃𝑃𝑃𝑖𝑖 𝑃𝑃𝑑𝑑 𝐺𝐺𝐺𝐺𝑃𝑃) + (% 𝐼𝐼𝑖𝑖𝑝𝑝𝑃𝑃𝑃𝑃𝑃𝑃𝑖𝑖 𝑃𝑃𝑑𝑑 𝐺𝐺𝐺𝐺𝑃𝑃) = 𝑃𝑃𝑑𝑑𝑔𝑔𝑃𝑃𝑑𝑑𝑑𝑑 𝑃𝑃𝑜𝑜 𝑃𝑃𝑝𝑝𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑖𝑖𝑖𝑖 This gives a value for the openness of each economy for every year in the database.
As measure of the “quality” of the labor force, the Human Capital Index as can be found in the Penn World Table is used. This measures human capital per person and uses data from Barro and
23
Lee (2013) this is then combined with a paper from Psacharopoulos (2004), which gives a certain rate of return on different years of schooling, the result is an index value, which is used as an indication of human capital per worker. The downside of this method is that it does not account for variations in the rates of return across countries and time.
Growth in ICT Capital deepening is calculated by taking data on ICT capital stock and
employment from the Conference Board’s Total Economy Database (TED) and the following formula:
% 𝐺𝐺𝑃𝑃𝑃𝑃𝑔𝑔𝑃𝑃ℎ 𝑃𝑃𝑑𝑑 𝐼𝐼𝑅𝑅𝐼𝐼 𝑃𝑃𝐿𝐿𝑝𝑝𝑃𝑃𝑃𝑃𝐿𝐿𝐶𝐶 𝑖𝑖𝑃𝑃𝑃𝑃𝑃𝑃𝑠𝑠 − % 𝐺𝐺𝑃𝑃𝑃𝑃𝑔𝑔𝑃𝑃ℎ 𝑃𝑃𝑑𝑑 𝑑𝑑𝑖𝑖𝑝𝑝𝐶𝐶𝑃𝑃𝑃𝑃𝑖𝑖𝑑𝑑𝑑𝑑𝑃𝑃 = 𝐺𝐺𝑃𝑃𝑃𝑃𝑔𝑔𝑃𝑃ℎ 𝑃𝑃𝐿𝐿𝑃𝑃𝑑𝑑 𝑃𝑃𝑜𝑜 𝐼𝐼𝑅𝑅𝐼𝐼 𝑃𝑃𝐿𝐿𝑝𝑝𝑃𝑃𝑃𝑃𝐿𝐿𝐶𝐶 𝑃𝑃𝑑𝑑𝑑𝑑𝑝𝑝𝑑𝑑𝑑𝑑𝑃𝑃𝑑𝑑𝑔𝑔
This gives ICT capital deepening and is basically a measure of the growth in the amount of ICT capital per worker.
Growth in intangible investments is calculated by taking growth rates of nominal intangible
investment stock data, as provided by the intangible investment database (www.intan-invest.net).
The data is somewhat limited as it only provides data for the year 1995 until 2010 and the data is also incomplete for some countries. This database also provides more detailed breakdowns of the data, making a distinction between computer software, innovative property and economic
competencies.
The Employment Protection Index is published by Eurostat and gives an indication of the amount of protection workers have against individual dismissals. There are three different versions available, but for this paper version 1 is used. The indicator includes eight different items regarding regulations, which are divided into three categories: “Procedural
Inconvenience”, “Notice and Severance Pay for No-Faults Individual Dismissal” and “Difficulty of Dismissal”.
Growth in capital deepening is calculated by using data on the growth in capital stock and growth of labor in the following formula:
% 𝐺𝐺𝑃𝑃𝑃𝑃𝑔𝑔𝑃𝑃ℎ 𝑃𝑃𝑑𝑑 𝑃𝑃𝐿𝐿𝑝𝑝𝑃𝑃𝑃𝑃𝐿𝐿𝐶𝐶 𝑆𝑆𝑃𝑃𝑃𝑃𝑃𝑃𝑠𝑠 − % 𝐺𝐺𝑃𝑃𝑃𝑃𝑔𝑔𝑃𝑃ℎ 𝑃𝑃𝑑𝑑 𝑑𝑑𝑖𝑖𝑝𝑝𝐶𝐶𝑃𝑃𝑃𝑃𝑖𝑖𝑑𝑑𝑑𝑑𝑃𝑃 = 𝐺𝐺𝑃𝑃𝑃𝑃𝑔𝑔𝑃𝑃ℎ 𝑃𝑃𝐿𝐿𝑃𝑃𝑑𝑑 𝑃𝑃𝑜𝑜 𝑃𝑃𝐿𝐿𝑝𝑝𝑃𝑃𝑃𝑃𝐿𝐿𝐶𝐶 𝑃𝑃𝑑𝑑𝑑𝑑𝑝𝑝𝑑𝑑𝑑𝑑𝑃𝑃𝑑𝑑𝑔𝑔 This gives the growth rate of the amount of capital per worker.
Data regarding the share of manufacturing in GDP is retrieved from the United Nations’ National Accounts Main Aggregates Database and calculated by dividing “Gross value added
manufacturing sector” by “Gross value added total economy”.
Domestic credit provided by financial sector (% of GDP) is gathered from the World Bank and is in growth rates.
24
In the following correlation matrix between all available variables, one can conclude that there are no problematic correlations (>0.8) between the independent variables that might lead to
(perfect) multicollinearity.
Correlation Matrix Sector Size Financial Manufac-Share of turing
Human Capital Index
ICT
Capital Investments Intangible
Compens-ation per Employee Openness Capital Deepening Financial Sector Size 1 Share of Manufacturing -0.2802 1 Human Capital Index -0.0591 -0.07 1 ICT Capital 0.111 -0.3218 0.2152 - 1 Intangible Investments 0.0872 0.0269 0.0049 0.3899 1 Compensation per Employee 0.0185 0.1238 0.0008 - 0.0492 - -0.1122 1 Openness -0.139 0.0316 0.2918 - 0.404 0.072 -0.1075 1 Capital Deepening 0.1672 -0.4669 0.0364 0.2312 -0.1751 -0.082 -0.1212 1 Employment Protection 0.0358 0.1507 0.0732 - 0.2394 - -0.3374 0.0641 0.2476 0.0286
25
5. Results
The data on human capital index, employment protection and openness of the economy, does not change much over time, however they do differ across countries. They are there for omitted from the regression and the resulting differences between countries are absorbed by the country
specific constants. Running this regression returns the following results:
Variables Model 1 Model 2 Model 3
Financial Sector Size 0.00148 (.0230) Share of Manufacturing 0.284*** 0.284*** 0.297*** (.0314) (.0291) (.0306) ICT Capital Deepening -0.0962* -0.0968* (.0496) (.0500) Real Intangible Investments 0.193*** 0.194*** 0.170*** (.0582) (.0556) (.0576) Real Compensation per Employee 0.148** 0.148** 0.171** (.0692) (.0684) (.0688) Capital Deepening 0.0392*** 0.0392*** 0.0374*** (.0072) (.0072) (.0069) Constant 0.354 0.361 -0.295 (.4390) (.4500) (.3190) Observations 150 150 150 R-squared 0.626 0.626 0.613
Robust standard errors in parentheses *** p<0.01, ** p<0.05, * p<0.1
Table 2, results of the regression
The first regression shows that share of manufacturing has a positive effect on labor productivity growth, meaning that a higher share of manufacturing in the economy means a higher labor productivity growth. ICT capital deepening has a negative effect, which is unexpected, as one would expect that investing in faster computers, for example, would increase labor productivity growth. There is no clear explanation for this result. Real intangible investments, real
compensation per employee and capital deepening all have positive coefficients, indicating that a rise in these variables increases labor productivity growth. Financial sector size is not statistically significant different from zero, therefor it is excluded in the second regression. As expected this does not change the values of the other coefficients by much. The third regression also excludes
26
ICT capital deepening, as it was only significant at a 10% level. This has somewhat of an effect on the variables, but all the signs remain the same en the values do not change dramatically.
The results show that the biggest effect is caused by changes in the share of manufacturing in GDP, with a coefficient of 0.297. The size of the effect of real intangible investments and real compensation per employee is almost the same (0.170 and 0.171 respectively). The size of the coefficient on changes in capital deepening is relatively small (0.0374). All of these coefficients are positive, so increases in these factors have a positive effect on labor productivity growth. This makes sense for capital deepening, as the more machines a worker has available, the more productive he will be. The positive coefficient on the share of manufacturing variable can be explained by the so-called “Baumol’s cost disease”, first introduced in a paper by Baumol and Bowen in 1966. It argues that while it is easy to increase productivity in manufacturing firms, either by increasing the number of machines or by using more advanced technology, this is not the case with service sectors firms. The latter is mostly dependent on human interactions and is therefore hard to replace by machines, this means that any increase in output must be due to an increase in the number of employees. While manufacturing sectors can increase productivity and thus increase the wages they pay, this is not the case with the service sector. They will still have to compete with the manufacturing sector for employees and must therefore offer higher wages as well.
The positive coefficient on real compensation per employee can be explained by the fact that workers that get a higher wage are more motivated and will work harder (also because they do not want to get fired and lose their relatively high wage).
27
However, these results are based on only 150 observations. This is because observations on real intangible investments are only available for 15 countries and 10 years. To increase the number of observations, real intangible investment can be excluded from the model. This gives 673 observations (33 countries with around 20 observations each, 1990-2012). The result of this regression is shown in table 3.
Variables Intangibles No Intangibles No Intangibles, full sample Share of Manufacturing 0.297*** 0.2988*** 0.223*** (0.0306) (0.0278) (0.0262) Real Intangible Investments 0.170*** - - (0.0576) - - Real Compensation per Employee 0.171** 0.2185*** 0.112*** (0.0688) (0.0497) (0.0428) Capital Deepening 0.0374*** 0.0372*** 0.0310*** (0.0069) (0.0067) (0.00404) Constant -0.295 0.6670*** 1.090*** (0.319) (0.1237) (0.122) Observations 150 225 673 R-squared 0.613 0.584 0.394
Robust standard errors in parentheses *** p<0.01, ** p<0.05, * p<0.1
Table 3, results from the standard regression
By excluding real intangible investments, the number of observations increases dramatically, from 150 to 673 but the R-squared decreases to 0.394. But this method is a combination of two separate steps. First, real intangible investments is excluded, this leads to 75 extra observations, because there are now five extra countries that had no information on this variable (so 5x15=75) and are now included in the sample. The variables change little, but the constant is now statistically significant.
The second step is the increase in years involved, as it was first limited to only 15 years per countries because of real intangible investments. One can see that the variables show a slight decrease compared to their original value, the signs, however, remain the same on all variables. So when increasing the number of observations by excluding real intangible investments, the remaining variables do not change much, we can therefor assume that these observations are
28
quite robust. Furthermore, we can observe that real intangible investments are important in
explaining the productivity puzzle, as excluding it decreases the R2 by around 0.2.
Another question is whether a model primarily based largely on pre-crisis data is still suitable to analyze a post-crisis phenomenon. We therefor test if the coefficients on pre-crisis data differ from post-crisis data. A simple method would be to cut the dataset in half, one with data up to 2007 and one with data from 2008 and onwards. But by doing this we would throw away some valuable information. A dummy variable is therefor created instead, given the value of 0 for years before 2008 and the value of 1 for the year 2008 and after. The resulting regression formula is then as follows:
(𝐿𝐿𝐿𝐿𝐿𝐿𝑃𝑃𝑃𝑃 𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃 𝑔𝑔𝑃𝑃𝑃𝑃𝑔𝑔𝑃𝑃ℎ)𝑂𝑂,𝑖𝑖
= 𝛼𝛼𝑖𝑖 + 𝛽𝛽1(∆𝑅𝑅𝐿𝐿𝑅𝑅)𝑂𝑂,𝑖𝑖+ 𝛽𝛽2(∆𝑃𝑃𝐿𝐿𝑝𝑝𝑃𝑃𝑃𝑃𝐿𝐿𝐶𝐶 𝑃𝑃𝑑𝑑𝑑𝑑𝑝𝑝𝑑𝑑𝑑𝑑𝑃𝑃𝑑𝑑𝑔𝑔)𝑂𝑂,𝑖𝑖
+ 𝛽𝛽3(∆𝑖𝑖ℎ𝐿𝐿𝑃𝑃𝑑𝑑 𝑃𝑃𝑜𝑜 𝑖𝑖𝐿𝐿𝑑𝑑𝑃𝑃𝑜𝑜𝐿𝐿𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑑𝑑𝑔𝑔 𝑃𝑃𝑑𝑑 𝐺𝐺𝐺𝐺𝑃𝑃)𝑂𝑂,𝑖𝑖+ 𝛽𝛽4𝐺𝐺(∆𝑅𝑅𝐿𝐿𝑅𝑅)𝑂𝑂,𝑖𝑖
+ 𝛽𝛽5𝐺𝐺(∆𝑃𝑃𝐿𝐿𝑝𝑝𝑃𝑃𝑃𝑃𝐿𝐿𝐶𝐶 𝑃𝑃𝑑𝑑𝑑𝑑𝑝𝑝𝑑𝑑𝑑𝑑𝑃𝑃𝑑𝑑𝑔𝑔)𝑂𝑂,𝑖𝑖+ 𝛽𝛽6𝐺𝐺(∆𝑖𝑖ℎ𝐿𝐿𝑃𝑃𝑑𝑑 𝑃𝑃𝑜𝑜 𝑖𝑖𝐿𝐿𝑑𝑑𝑃𝑃𝑜𝑜𝐿𝐿𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑑𝑑𝑔𝑔 𝑃𝑃𝑑𝑑 𝐺𝐺𝐺𝐺𝑃𝑃)𝑂𝑂,𝑖𝑖
+ 𝜀𝜀𝑂𝑂,𝑖𝑖
Where coefficients 𝛽𝛽4, 𝛽𝛽5and 𝛽𝛽6 show the change in the respective variable when data from the
years 2008-2012 is used. This gives the following results: Variables
Share of Manufacturing 0.2003***
(0.0262)
Real Compensation per Employee 0.0825***
(0.0428)
Capital Deepening 0.0446***
(0.00404)
Share of Manufacturing post-crisis -0.0062
(0.0441) Real Compensation per Employee
post crisis
-0.0007 (0.0924)
Capital Deepening post crisis -0.0335***
(0.0066)
Constant 1.0254***
(0.122)
Observations 673
R-squared 0.4490
Robust standard errors in parentheses *** p<0.01, ** p<0.05, * p<0.1
29
This shows that for post-crisis data the coefficient of capital deepening is statistically significant different from its pre-crisis value. The pre-crisis value of this coefficient can be easily read from the table (0.0446), to get the post crisis coefficient the other coefficient is added to this value: 0.0446–0.0335=0.0111. This means a significant lower effect of capital deepening after the crisis. This might be due to the crisis, as during a recession investment in capital may be harder for firms as banks are more reluctant to give credit. Both share of manufacturing and real compensation per employee did not statistically significant change compared to the pre-crisis value. This shows that the effect of share of manufacturing has remained about stable throughout the sample period.
For some of the included countries there are some missing data points, a number of eastern European countries have this problem for the first few years after 1990 for example, also Estonia and Chile are missing date over the whole sample, so the following results are from a regression that excludes Chile, Estonia and only involves data from 1995-2012. It involves 546 observations, this is less than one would expect, as there are 31 countries and 17 years giving 527 expected observations. This is due to the following missing data points:
• Growth in labor productivity of Slovenia in 1996
• Growth in real compensation per employee in Greece for 1996-2000 • Growth of capital deepening in Iceland for 1996 and 1997
This total of 8 missing data points is the difference between the expected and actual number of observations (527-519=8).
30 Variables Coefficients Share of Manufacturing 0.2056*** (0.0274) Real Compensation per Employee 0.1282** (0.0472) Capital Deepening 0.0327*** (0.0045) Constant 1.0301*** (0.1155) Observations R-squared 546 0.4136 Robust standard errors in parentheses
*** p<0.01, ** p<0.05, * p<0.1
Table 5, results from regression with no missing data points
These results still look very similar to the ones we had before. The results are therefor not influenced by countries with large amounts of incomplete data.
Combining the coefficients from the last column of table 3 with the values from the database gives the following graph:
Graph 16, Labor productivity growth implied by the model, compared to actual value
-3 -2 -1 0 1 2 3 4 5 6 7 Gro w th in %
Labor productivity growth in the UK
1998-2012
implied actual
31
It shows that our model underestimates labor productivity growth in the pre-crisis era. It also shows that it captures the decline in labor productivity growth around 2010 very well, explaining a large portion of the shown decline.
Implied change in growth Actual change in growth 2.56% 2.97% percentage explained 86.04%
Table 6, the model’s implied change in growth compared to the actual value, as percentage
For the years 2010, 2011 and 2012, the variable “share of manufacturing” is responsible for between 30% and 50% of the implied change in growth in the model. With the remainder being divided across the other variables.
This leads us to conclude that the decline in labor productivity growth in OECD countries over the period after 2008 was mainly driven by changes in the share of manufacturing firms in GDP, growth rate of capital deepening, intangible investments growth rate and the growth in real labor costs. Of these four factors, share of manufacturing in GDP is the prime suspect, as this is the only variable in our analysis that clearly differs from other OECD countries. We can therefor conclude that share of manufacturing in GDP is in this analysis the factor that explains (part) of the labor productivity puzzle in the UK. A number of variations of the model have been tested, with various assumptions, but the results remained the same throughout, proving that these results are also very robust.
32
6. Summary
In short, the puzzle that we are facing is a decline in labor productivity growth in the UK since the start of the 2008 financial crisis, that is not observed in any other developed country. In the current literature and research there is no consensus on what is exactly causing this decline in labor productivity growth. But a number of theories have been proposed: labor hoarding, impaired reallocation of capital, lower capital-labor ratio, less flexible labor market, intangible investments and a shift in the structure of the economy. This paper tries to find out which of these theories are the best in explaining the decline in labor productivity growth in the UK.
The first step in this paper is to see which factors actually influence labor productivity growth in general. Using a time series regression on data from 33 OECD countries, with a timespan of 1990 until 2013, four statistically significant factors were identified. They are: share of manufacturing in GDP, real intangible investments, real compensation per employee and capital deepening.
Because data on real intangible investments is very limited, the total number of observations was only 150. When excluding real intangible investments, the total number of observations increased to 673 and the coefficients for the other variables did not change much. From the various robustness tests it also appears as if the results are reliable.
Of these four variables, share of manufacturing in GDP is the most interesting, as it was the only variable that behaved differently in the UK compared to the other OECD countries. It is also the only country where there is a clear downward trend in the share of manufacturing in GDP visible over time that remains constant after 2009. This is combined with the fact that up to around 2010, service sector firms were more productive than manufacturing sector firms. It therefor seems that the trend to move away from manufacturing firms, as service sector firms were more productive, has been halted and that the UK’s economy lost an important driver of labor productivity growth that might explain the poor labor productivity growth of recent years.
33
7. Literature list
Aghion, P., Bloom, N., Blundell, R., Griffiths, R. & Howitt, P., (2001). Empirical Estimates of
the Relationship between Product Market Competition and Innovation, Quarterly Journal of
Economics 120, 701–728.
Aghion, P. T. B., Browne, J., Caselli, F., Lambert, R., Lomax, R., Stern, N. & Van Reenen, J., (2013). Investing for Prosperity. Report of the LSE Growth Commission.
Van Ark, B., Inklaar, R., & McGuckin, R. H. (2003). ICT and Productivity in Europe and the
United States Where do the differences come from?. CESifo Economic Studies, 49(3), 295-318.
Aschauer, D. A., (1989). Is public expenditure productive? Journal of monetary economics, 23(2), 177-200.
Barro, R. J., & Lee, J. W. (2013). A new data set of educational attainment in the world, 1950–
2010. Journal of development economics, 104, 184-198.
Bartelsman, E., Haltiwanger, J. & Scarpetta, S., (2013). Cross-country differences in
productivity: The role of allocation and selection. The American Economic Review, 103(1), 305-334.
Blundell, R., Crawford, C. & Jin, W., (2013). What Can Wages and Employment Tell Us About the UK's Productivity Puzzle. Institute for Fiscal Studies (IFS), Working Paper, (W13/11). Bloom, N., Van Reenen, J., & Sadun, R., (2005). It Ain’t What You Do, it’s the Way that You do
IT. London School of Economics.
Broadbent, B., (2012). Productivity and the allocation of resources. (Speech given to the Durham Business School)
Criscuolo, C., Haskel, J. & Slaughter, M., (2004). Why Are Some Firms More Innovative?
Knowledge Inputs, Knowledge Stocks and the Role of Global Engagement, NBER.
Criscuolo C. & Martin R., (2003). Multinationals and US Productivity Leadership, Evidence
from Britain. Draft note, Centre for Research into Business Activity.
Disney, R., Haskel, J. & Heden, Y., (2003). Restructuring and Productivity Growth in UK
34
Fernald, J. G., (1999). Roads to prosperity? Assessing the link between public capital and productivity. American Economic Review, 619-638.
Field, S. & Franklin, M., (2013). Micro-data Perspectives on the UK Productivity Conundrum-An Update. Office of National Statistics.
Grice, J., (2012). The Productivity Conundrum, Interpreting the Recent Behaviour of the
Economy. Office of National Statistics.
Haskel, J., Cereda, M., Crespi, G. & Criscuolo, C., (2005). Creativity and Design Study. Economics Paper No.15, Department of Trade & Industry.
Haskel, J., Goodridge, P., & Wallis, G., (2013). Can intangible investment explain the UK
productivity puzzle? National Institute Economic Review, vol. 224.
Haskel, J., Hawkes, D. & Perriera, S., (2004). How much do Skills Raise Productivity? UK
Evidence from Matched Plant, Worker and Workforce Data. Centre for Research into Business
Activity.
Haskel J., & Khawaja N., (2003). Productivity in UK Retailing, Evidence from UK Microdata. Discussion Paper, Centre for Research into Business Activity.
Holtz-Eakin, D., (1992). Public-sector capital and the productivity puzzle. Working paper No. 4122, National bureau of economic research.
Lynde, C., & Richmond, J., (1993). Public capital and total factor productivity. International Economic Review, 401-414.
King, S., (2013). Putting lower employment ahead of higher productivity risks long-term
economic damage. The Times, March 19th 2013.
Marrano, M. G., Haskel, J., & Wallis, G., (2009). What happened to the knowledge economy? ICT, intangible investment, and Britain's productivity record revisited. Review of Income and Wealth, 55(3), 686-716.
Martin, B., & Rowthorn, R., (2012). Is the British economy supply constrained II? A renewed critique of productivity pessimism. Centre for Business Research, University of Cambridge. Munnell, A. H., (1990). Why has productivity growth declined? Productivity and public investment. New England Economic Review, (Jan), 3-22.
35
Oulton, N., Sebastia-Barriel, M., (2013). Long and short-term effects of the financial crisis on
labour productivity, capital and output. Working Paper no. 470, Bank of England.
Patterson, P., (2012). The productivity conundrum, explanations and preliminary analysis. Office of National Statistics.
Pessoa, J. P., & van Reenen, J. M., (2013). The UK productivity and jobs puzzle: does the answer lie in labour market flexibility? Centre for Economic Performance, LSE.
Psacharopoulos, G., & Patrinos, H. A. (2004). Returns to investment in education: a further
update. Education economics, 12(2), 111-134
De Serres, A., Yashiro, N. & Boulhol, H., (2014). An international Perspective on the New
Zealand Productivity Paradox. OECD working paper
Sondermann, D., (2012). Productivity in the euro area, any evidence of convergence? European Central Bank, Working Paper No. 1431
Timmer, M., Inklaar, R., O'Mahony, M. and van Ark, B., (2010). Economic growth in Europe: a comparative industry perspective (Vol. 1). Cambridge University Press.
