Decomposing the Productivity- Wage Nexus in Selected OECD Countries,

Transcription

1 Decomposing the Prouctivity- Wage Nexus in Selecte OECD Countries, Anrew Sharpe Centre for the Stuy of Living Stanars James Uguccioni Centre for the Stuy of Living Stanars 1 ABSTRACT Stanar economic theory preicts that in the long run, prouctivity growth ought to rive aggregate real wage growth. We consier this preiction in the case of 11 OECD countries, an fin that eight of the 11 experience slower meian real wage growth than labour prouctivity growth over the perio. We ecompose the gap between labour prouctivity growth an meian real wage growth into four components: wage inequality, changes in the importance of employer contributions to social insurance programs, ifferences between the prices of output an consumption, an changes to labour's share of income. The ecompositions ultimately show that there is no common cause for the prouctivity-wage gap, though most countries i see wage inequality grow an labour's share of income fall to some egree over our perio of stuy. In the face of growing inequality in avance economies, the OECD (2014) has initiate a significant research effort aime at unerstaning an promoting inclusive growth. The aim is to avance policies to ensure that the benefits of growth are broaly share. Across OECD countries, governments are searching for ways to ensure that subsets of society are not left behin by economic growth. For example, the Canaian government has installe a Cabinet Committee on Inclusive Growth, Opportunities an Innovation with the manate to "[consier] strategies esigne to promote inclusive economic growth, opportunity, employment an social security" in Canaa. These efforts are timely because evience on wage growth suggests that economic growth has not been broaly share in recent ecaes. In eight of the 11 OECD countries examine in this article, meian real wage growth since the mi-1980s has not kept pace with labour prouctivity growth. The size of the growth gap between labour prouctivity an meian real wages iffers across countries, but the qualitative pattern is consistent: workers are growing more prouctive, but those prouctivity gains are not being matche by growth in the typical worker's wage. Economic history an economic theory suggest that labour prouctivity growth shoul 1 Anrew Sharpe is the Excecutive Director at the Centre for the Stuy of Living Stanars (CSLS) James Uguccioni was an economist at the CSLS at the time of writing. The authors woul like to thank Guiseppe Nicoletti an two anonymous referees for useful comments. This is an abrige an revise version of Uguccioni an Sharpe (2016). The paper was first presente at the OECD's First Global Forum on Prouctivity hel in Lisbon, Portugal July 7-8, 2016: anrew.sharpe@csls.ca. I N T E R N A T I O N A L P R O D U C T I V I T Y MON I T O R 25

2 generate rising living stanars for workers over time, so the apparent isconnect between labour prouctivity growth an wage growth is puzzling. What factors account for it? In this article, we show that the gap between labour prouctivity growth an meian hourly earnings growth can be ecompose into contributions from the following four sources: rising earnings inequality; changes in the importance of employer contributions to social insurance programs as a form of labour compensation; rising relative prices for consumer goos; an a ecline in labour's share of aggregate income. Each of these components has its own implications for the welfare of workers. To the extent that the prouctivity-earnings gap simply reflects a rising share of labour compensation being pai in the form of employer contributions to social insurance plans, for example, it is not obvious that workers are any worse off. On the other han, rising earnings inequality or a ecline in labour's share of income might represent more serious obstacles to broa-base prosperity. We perform the ecomposition for 11 OECD countries: Canaa, Denmark, France, Finlan, Germany, Irelan, the Netherlans, Norway, Spain, the Unite Kingom, an the Unite States. The ecompositions show that the prouctivity-wage growth gap has no single common cause across the countries, but most countries i experience rising earnings inequality an a ecline in labour's share of income over our perio of stuy. The ecompositions typically run from the mi to late 1980s through to 2010 or 2013, epening on the availability of househol survey ata for a given country. The article is comprise of five sections. The first iscusses literature that provies context for our analysis. The secon section escribes our framework for ecomposing the gap between labour prouctivity growth an meian real hourly wage growth. Section three presents the results of the ecomposition. The fourth section iscusses wage growth throughout the wage istribution in more etail. Section five conclues. Literature Review The failure of real wages to keep pace with labour prouctivity is not a new observation. Fisher an Hostlan (2002) observe that labour prouctivity outstrippe real wage growth in Canaa from 1994 to Bartlett an Tapp (2012) foun that labour prouctivity growth outpace labour compensation growth from the mi-1990s through to 2012 in Canaa. The gap, however, is not limite to Canaa. The International Labour Organization (2015) observe that labour prouctivity growth exceee real wage growth from 1999 to 2013 in evelope countries across the boar. Decompositions allow analysts to ientify the proximate sources of the gap between labour prouctivity growth an real wage growth. In a stuy of the American non-farm business sector from 1970 to 2006, Felstein (2008) foun that average real wage growth was inee lower than labour prouctivity growth. The ifference was a matter of prices. When he ajuste wages for inflation using the non-farm business sector output price inex (rather than the consumer price inex), he foun that wages grew at approximately the same rate as labour prouctivity. For research that relates the growth of wages an labour prouctivity, Felstein stresses the importance of accounting for ifferences in price inexes an the importance of using total compensation (i.e. incluing supplementary labour income an fringe benefits) instea of only wages an salaries when calculating a wage for comparison with labour prouctivity. We 26 N U M B E R 32, S P R I N G 2017

3 hee both of Felstein's concerns in our analysis. While Felstein's ecomposition provies a framework for relating labour prouctivity growth to average wage growth, he fails to consier how wage growth was actually experience by the workers near the meian - a better measure of the wage of the typical 'mile class' worker. Sharpe et al. (2008a; 2008b) consier how wage growth was experience by the meian worker, ecomposing the gap between labour prouctivity growth an real meian wage growth in Canaa into four contributing factors: rising inequality, poor terms of trae for labour, a ecrease in labour's share of income, an measurement inconsistencies. 2 They fin that from 1980 to 2005, labour prouctivity grew 1.26 percentage points per year faster than meian real earnings. They ecompose the gap into their four factors, attributing 0.35 percentage points per year to inequality, 0.42 percentage points per year to terms of trae for labour, 0.25 percentage points per year to labour's share of income, an 0.25 percentage points per year to measurement issues. This report follows the metho of Sharpe et al. but extens the analysis to ten aitional OECD countries. Pessoa an Van Reenen (2012) perform a ecomposition of meian wage growth an prouctivity growth similar to the one presente in Sharpe et al. (2008b) for the Unite Kingom an the Unite States. They propose that there are two ifferent types of measurements for the ivergence - "gross ecoupling" an "net ecoupling". The former measures ifferences in growth between labour prouctivity an meian hourly real earnings, while the latter measures ifferences in growth between labour prouctivity an average labour compensation per hour (eflate with the same eflator). Gross ecoupling accounts for changes to labour's share of income, labour's terms of trae, changes meian an mean hourly earnings, an the wege between labour compensation an earnings, while net ecoupling only accounts for changes to labour's share of income. Ultimately, Pessoa an Van Reenen (2012) fin little evience of net ecoupling in the UK, but significant gross ecoupling in the Unite States an the UK. In the UK, gross ecoupling was riven by ifferences between mean an meian earnings an the wege between earnings an labour compensation. Pessoa an Van Reenen (2012) recognize that both gross ecoupling an net ecoupling are important policy inicators. Gross ecoupling relates the "true mile" of the earnings istribution to labour prouctivity. It also eflates earnings with the CPI an labour prouctivity with the GDP eflator, capturing any ifference in the prices face by firms an workers. This is an important istinction to make because firms an consumers can at times face very ifferent prices. Changes in capital equipment prices affect firms more than consumers, for example. Net ecoupling, on the other han, is important because it challenges one of the main stylize facts cite by economists - labour's stable share of income. Pessoa an Van Reenen observe that net ecoupling coul occur for many reasons, incluing shocks which isturb the long run equilibrium, technological bias against labour, changes to the level of competition in the market (in the prouct market it results in setting higher prices, while in the labour market it results in setting lower wages), an finally changes to labour supply ue to structural phenomena like globalization. Mishel an Gee (2012) use the methoology evelope by Sharpe et al (2008b) to compare 2 The term "measurement inconsistencies" refers to the combine effect of employer social contributions an changes in hours of work per worker. I N T E R N A T I O N A L P R O D U C T I V I T Y MON I T O R 27

4 the growth of meian real wage in the Unite States with labour prouctivity. Like most of the literature, they also fin a significant gap between growth in labour prouctivity an meian real wages: 1.56 percentage points between 1973 an Rising wage inequality accounte for 0.61 percentage points, while labour's terms of trae accounte for another 0.44 percentage points. They specifically point to the erosion of labour stanars, globalization, high trae eficits, an the rising share of capital epreciation in GDP to explain both growing inequality an the changes in the istribution of income towars capital. A recent OECD stuy by Schwellnus et al. (2017) provies an analysis of the ecoupling of meian wages from prouctivity in OECD countries for the perio base on trens in labour s share an the ratio of meian to average wages. It fins that labour prouctivity grew faster than meian wages in 15 of 24 countries. Empirical Framework Our ecomposition of the gap between labour prouctivity growth an meian real hourly earnings growth follows the approach evelope in Sharpe et al. (2008a). In this section, we formally escribe this approach. Decomposition Metho The starting point for the ecomposition is the following accounting ientity: Y Y Y P L = L L Y P L C P L Y Y P C (1) Here, Y L is total nominal labour compensation, P C is the price of consumption goos, an L is total hours worke. Y is total nominal output (or income) in the economy an P Y is the price of output. Thus, the ratio Y L ( P C L) enotes average real hourly labour compensation in units of consumption goos (i.e. the "consumer wage"). On the right-han sie, the ratio Y ( P Y L) enotes real output per hour in units of output goos; that is, labour prouctivity. is labour's share of total income in the economy. The remaining term P Y P C is the relative price of output goos in terms of consumption goos; following the literature, we will refer to this as "labour's terms of trae." For any variable X, let the notation %X enote the per cent growth rate of X. Then expressing equation (1) in growth rates, we obtain: %Average Real Hourly Compensation = % Labour Prouctivity + % Labour Share % Labour Terms of Trae (2) Our goal is to explain changes in the gap between labour prouctivity an meian real hourly earnings. Let % Gap enote the prouctivity-earnings growth gap. Formally, it is efine by: % Gap = % Labour Prouctivity (3) - % Meian Real Hourly Earnings Rearranging (2) an using (3) to eliminate labour prouctivity growth, we obtain: % Gap = % Average Real Hourly Compensation - (4) % Meian Real Hourly Earnings - % Labour Share- % Labour Terms of Trae Now, the change in average real hourly earnings relative to meian real hourly earnings is an inicator of the change in earnings inequality over time. Thus, we efine the change in inequality as: % Inequality = Y L Y % Average Real Hourly Earnings - (5) % Meian Real Hourly Earnings Finally, we nee to relate average real hourly compensation to average real hourly earnings. 28 N U M B E R 32, S P R I N G 2017

5 The ifference between these two measures reflects the impact of changes in employer contributions to social insurance programs: % Average Real Hourly Compensation - % Average Real Hourly Earnings = (6) % Employer Social Contributions Substituting (5) an (6) into (4) yiels the overall ecomposition: % Gap = % Inequality + % Employer Social Contributions - (7) % Labour Terms of Trae - % Labour Share equation (7) is the final ecomposition formula. We fin the accounting approach very useful. It raws our attention to the relationships between the prouctivity earnings gap an several other economic phenomena such as: rising earnings inequality, the changing impact of laws governing employer contributions to social insurance programs. It lens a iscipline, quantitative characterization to those relationships. It suggests areas for future research that might clarify the causal mechanisms at play. The ecomposition in equation (7) oes not, on its own, justify any statements about cause an effect an oes not explain the trens observe. To aress such questions woul require a structural moel that explains why each of the components change the way it i. Data Sources Our analysis relies on two ata sources: national accounts an househol surveys. 3 For estimates base on national accounts ata, we employ the OECD National Accounts from the OECD Stat public-use atabase. For estimates that rely on househol surveys (meian an average earnings from househol surveys), we rely on the micro-atasets mae available by the Luxembourg Income Stuy. Table 1 etails the specific survey(s) use for each country. The length of our time series varies by country with househol survey availability. Generally, the series span from 1986 or 1987 to 2010 or Germany an Irelan are the two exceptions to the rule, with our time series for the two countries spanning 1994 to To create our meian an average wage series for each country, we use the annual labour income for both part-time an full-time employees from the relevant househol survey. We exclue self-employe from our sample when generating the istribution of annual labour income in a given country because of ata issues in ifferentiating labour income from returns to capital. 5 In orer to create average hourly real wage an meian hourly real wage estimates, we then ivie through by the average hours worke per person employe an eflate each series with the CPI. 6 3 The ata series use in this article can be foun at 4 Irelan began in 1994 simply ue to ata availability. We opte to begin our German series in 1994 because it was the first househol survey after East an West Germany were reunite, an we lack microata from East Germany prior to the Wall coming own. 5 The primary ifficulty with self-employe ata is that their annual income comes both from the labour the self-employe put in their business an the return on the capital they have investe. Most countries have tax systems set up in such a way that iviens from a business are treate ifferently than salaries pai out from the business. As such, the self-employe will naturally take into account tax implications when eciing how they will be remunerate in a given year. By excluing the self-employe, we avoi any changes to labour income which are the result of changes to the tax treatment of iviens. Moreover, as our ecomposition is an exercise in growth, so long as "true" self-employe labour income i not grow faster or slower than labour income i for employees, we o not lose any information by ropping the self-employe. I N T E R N A T I O N A L P R O D U C T I V I T Y MON I T O R 29

6 Table 1: Househol Survey Microata Sources Country Survey(s) Use by LIS Canaa Survey of Consumer Finance (1987, 1991, 1994, 1997), Survey of Labour an Income Dynamics (1998, 2000, 2004, 2007, 2010) Denmark Law Moel (1987, 1992, 1995, 2000, 2004, 2007, 2010) Finlan Income Distribution Survey (1987, 1991, 1995, 2000, 2004), Survey on Income an Living Conitions (2007, 2010, 2013) France Family Buget Survey (1984, 1989, 1994, 2000, 2005, 2010) Germany German Social Economic Panel Stuy (1994, 2000, 2004, 2007, 2010) Irelan Living in Irelan Survey (1994, 1995, 1996, 2000), Survey on Income an Living Conitions (2004, 2007, 2010) Aitional Enquiry on the Use of (Public) Services (1983, 1987, 1990), Socio- Netherlans Economic Panel Survey (1993, 1999), Survey on Income an Living Conitions (2004, 2007, 2010) Norway Income Distribution Survey (1986, 1991, 1995, 2000, 2004), Househol Income Statistics (2007, 2010) Family Expeniture Survey (1980, 1990), Spanish European Community Spain Househol Panel (1995, 2000), Survey on Income an Living Conitions (2004, 2007, 2010, 2013) Unite Kingom Family Expeniture Survey (1986, 1991, 1995), Family Resources Survey (1994, 1999, 2004, 2007, 2010, 2013) Current Population Survey March Supplement (1986, 1991, 1994, 1997, Unite States 2000), Current Population Survey Annual Social an Economic Supplement (2004, 2007, 2010, 2013) Source: Luxembourg Income Stuy Decomposition Results This section presents an iscusses the ecomposition results. We begin with an overall summary of the results. We then evote one subsection to etaile analysis of each of the four components: earnings inequality, employer social contributions, labour's terms of trae, an labour's share of income. Summary of Results The ecomposition results are summarize in Table 2. Overall, eight out of the 11 OECD countries stuie saw labour prouctivity grow faster than meian real hourly wages (Chart 1). The gap was largest in the Unite States, at 1.47 per cent per year from 1986 to On the other en of the spectrum, Spain, Norway, an Irelan all experience faster meian hourly real wage growth than labour prouctivity growth, resulting in a shrinking prouctivity-wage gap over their respective time perios. The importance of the four components of the gap varie significantly by country. In Canaa an the Unite Kingom, rising inequality was the largest contributor to the gap. In Germany, the Unite States, an Norway, labour's terms of trae ha the largest absolute effect on the gap. In Finlan an the Netherlans, labour's falling share of income was the largest contributor to the gap. The size of a component of the gap within a country can give some inication to policymakers where action may nee to be taken to reuce the prouctivity-wage gap. 6 Amittely, using average hours worke in an economy to generate an hourly wage series from the micro-ata is not ieal. Ieally, the househol surveys woul also inclue a weekly or annual hours worke variable, from which we coul create hourly wage (more recent surveys o ten to inclue such variables, but changes over short perios are less informative for prouctivity research). However, as average hours worke is riven by full-time workers, we can interpret the general ecline of average hours worke as a representative tren for all full-time workers. As our ecomposition eals in growth rates rather than levels, our use of average hours worke to generate hourly wages shoul not introuce bias into our results, particularly for wages levels in the mile of the istribution (i.e. meian an average). Bick et al. (2016) present a more etaile breakown of the ecline of hours across high income countries. 30 N U M B E R 32, S P R I N G 2017

7 Table 2: Decomposition of the Gap between Labour Prouctivity an Meian Real Hourly Earnings Growth into Four Components, Selecte OECD Countries, (average annual rate of change) > W DZ, ' ' / ^ > >^ WW' h^ e e e e ' e e e e & e ee e e e e * e ee e ee e * e e e e h< e e e e E e ee e e e e & e e ^ e e e E e e e e e e / e e ee e W' h^ e e e ' e e e & ee e * e ee * e e h< e e e E e & ee e e ee ^ ee e e ee E ee e e / e eee e ee Note: * , , , All others are Source: CSLS calculations from OECD National Accounts ata an househol survey microata from the Luxembourg income Stuy: Chart 1: Gap between Labour Prouctivity an Meian Real Hourly Wages Growth, Selecte OECD Countries, (percentage points per year) W Note: * , , , All others are Source: Table 2 I N T E R N A T I O N A L P R O D U C T I V I T Y MON I T O R 31

8 Table 3: Average an Meian Real Hourly Earnings, Selecte OECD Countries, (average annual rate of change) Z, DZ, / l h^ ee ' e & e ee e * ee ee * e e e h< e e e E e ee e & e e ^ e E e / ee ee Note: * , , , All others are Source: Househol Survey Microata from the Luxembourg Income Survey Chart 2: Ratio of Average to Meian Hourly Real Wage, Selecte OECD Countries, 2013 e Z e e D D Note: *2010 Source: Househol Survey Microata from the Luxembourg Income Survey Wage Inequality The wage inequality component is the gap between the growth rates of average an meian real hourly earnings. Empirically, earnings istributions within OECD countries are positively skewe; the mean is greater than the meian because the mean is ragge upwar by very high earners. When earnings at the top of the istribution grow more quickly than those in the mile of the istribution, the mean rises relative to the meian an earnings inequality rises. This woul imply that the gains from labour prouctivity are flowing isproportionately to workers who were alreay high earners relative to the meian worker, so % Inequality contributes positively to % Gap. The 11 OECD countries in our sample ha ifferent experiences with inequality growth over their respective perios. Generally in line with the wage inequality literature, most coun- 32 N U M B E R 32, S P R I N G 2017

9 Chart 3: Inequality Component, Percentage Point Contribution to the Gap, (per year) Note: * , , , tries experience rising inequality in recent ecaes accoring to our measure. As shown in Table 3, only France saw wage inequality fall overall, though meian hourly real wage growth only outpace average hourly real wage growth by 0.06 percentage points per year. As Chart 2 emonstrates, the level of wage inequality also varie significantly across countries: in 2013 in the Unite States the average real hourly wage was per cent of the meian hourly real wage, while in 2010 in Denmark the proportion was only per cent. The level of wage inequality in a country is very much the result of how the meian an mean have grown relative to one another over time. Chart 3 illustrates the percentage-point contributions of the wage inequality component to the gap in the 11 OECD countries. Inequality mae the largest contribution in Irelan, where the average hourly real wage grew faster than the meian hourly real wage by 0.88 percentage points per year. Inequality mae large contributions to the gap in both the Unite States an the Unite Kingom as well, contributing 0.52 an 0.49 percentage points per year, respectively. While evaluating the absolute percentage point contribution of equality to a country's overall gap is important, Table 2 as the imension of what proportion of a country's gap is ue to inequality. For example, espite inequality in Irelan making a large positive contribution to the gap, it was more than offset by the other three contributors. Contrarily, in the Netherlans an Canaa inequality contribute more than 50 per cent of the gap, an in the Unite Kingom it accounte for more than 100 per cent of the gap. Overall, there is no oubt that wage inequality has been growing across the OECD for ecaes. In most cases, the average hourly real wage grew aroun 0.10 to 0.50 percentage points per year faster than the meian hourly real wage equivalent to somewhere between 2 an 10 percentage points more cumulative growth over a 20 year perio. Eviently, these minor ifferences in growth can have major ramifications on the overall income istribution in the long run. It is, however, important to bear in min that ifferences in growth between the meian an the mean may fail to capture some important changes in the earnings istribution. In Section V, we iscuss alternative measures of inequality to learn about wage growth throughout the wage istribution. I N T E R N A T I O N A L P R O D U C T I V I T Y MON I T O R 33

10 Chart 4: Employer Social Contributions Component, Percentage Point Contribution to the Gap, Selecte OECD Countries, (per year) Note: * , , , Source: Table 2 Employer social contributions In principle, the ifference between average hourly earnings an average total labour compensation is that the latter captures employer social contributions (also calle supplementary labour income) while the former may not. 7 It is possible that part of the gap between labour prouctivity growth an meian hourly earnings growth is accounte for by workers receiving a growing share of their compensation in the form of employer contributions to social insurance programs rather than cash or in-kin earnings. 8 Whether this makes workers worse off epens on how much they value the social programs. Employer social contributions as a share of labour compensation have been growing throughout the OECD over recent ecaes. In Canaa, for example, employer social contributions as a share of labour compensation grew by about five percentage points from 1987 to This means that employer social contributions grew about 1.76 percentage points per year faster than wages an salaries over the perio (Uguccioni, Murray an Sharpe, 2016). In practice, we raw average hourly earnings from househol surveys an average hourly labour compensation from the National Accounts. We believe that employer social contributions are the main source of the growth iscrepancy between the two series (an that is why we have name this component of the gap 'employer social contributions'), but it is likely that other measurement iscrepancies between the two ata sources are capture here as well. The efinitions of labour income use in househol surveys may iffer across countries in subtle but important ways (e.g. in their treatment of bonuses or of non-cash income such as stock options). Sampling error in the surveys is another potential source of measurement iscrepancies. (It is well known, for example, that the top per cent of earners is ifficult to capture in househol surveys). 7 Supplementary labour income inclues contributions employers make on behalf of employees to state-run schemes such as national pension plans, unemployment insurance, an workplace injury insurance, as well as health an ental insurance plans provie by the employer, sickness an life insurance, an retirement allowances. 8 It can be note that efinitional ifferences between the ata sources for earnings an labour compensation, an changes in these ifferences over time, may also lea to ifferent growth rates for earnings an labour compensation. 34 N U M B E R 32, S P R I N G 2017

11 Chart 5: Labour's Terms of Trae, Percentage Point Contribution Per Year to the Gap, Selecte OECD Countries, (per year) Note: * , , , If no perio is note, the perio is Source: Table 2 As shown in Chart 4, this component's contribution to the gap in Irelan, France, an Den- bunle of goos consume by consumers is not average prices iffer because, in general, the mark exceee 0.50 percentage points per year the same as the bunle of goos prouce in the in absolute value. This inicates that there are omestic economy. 9 significant ifferences between the labour compensation component of the SNA an the hourly of output prouce per hour of work, so the rel- Labour prouctivity is efine as the volume earnings from the househol surveys prouce evant price is P Y. Workers ultimately want to use by the national statistics agencies in these countries, but further research is neee before a relevant price for measuring real labour com- their wages to buy consumption goos, so the efinitive conclusion is reache. pensation is P C. The iscrepancy between labour In per cent terms, employer social contributions (plus other measurement iscrepancies) therefore influence by the ratio P Y /P C. Follow- prouctivity an real labour compensation is make consierable contributions to the gap in ing the literature, we refer to this ratio as Irelan, Denmark, an France (Table 2). In Finlan, Norway, an Spain employer social contri- When % Labour Terms of Trae > 0, con- "labour's terms of trae." 10 butions make up a large share of the gap in sumer prices are falling relative to output prices. relative terms. Everything else being equal, this increases workers' purchasing power relative to labour Labour's terms of trae prouctivity, an hence reuces the gap between The accounting ientity in equation (1) labour prouctivity growth an real earnings inclues two prices: the consumption goos growth. That is why labour's terms of trae price P C an the output goos price P Y. These enter equation (7) with a negative sign. 9 For example, countries prouce goos that are exporte to other countries rather than purchase by omestic consumers. The prices of those exports are inclue in the output price P Y but not in the consumer price P C. 10 Clearly, an analogy is being rawn between P Y /P C an the more common notion of "terms of trae," which is the ratio of a country's export prices to its import prices. Intuitively, P C is the price of the goos workers buy an P Y is the price of the goos workers prouce. It is to workers' avantage when the price of what they sell increases relative to the price of what they buy, just as it is to a country's avantage when the price of what it sells (its exports) increases relative to the price of what it buys (its imports). I N T E R N A T I O N A L P R O D U C T I V I T Y MON I T O R 35

12 Chart 6: CPI an GDP Deflator Growth, Per cent Per Year, Selecte OECD Countries, Source: OECD CPI GDP Deflator Labour's terms of trae mae a sizeable contribution to the gap in six of the 11 countries (Chart 5). Labour's terms of trae in Norway contribute percentage points per year. Norway was the sole country where the GDP eflator outpace the CPI by such a wie margin (3.96 per cent per year versus 2.80 per cent per year) (Chart 6). This is explaine by much faster growth in export prices than consumption prices, riven by the commoity boom an large share of offshore oil an gas prouction in GDP. Germany an the Unite States ha the opposite experience than Norway i with labour's terms of trae. The two countries respectively saw the CPI grow 0.59 percentage points an 0.57 percentage points faster than the GDP eflator. In the Unite States, the relatively high rate of growth sustaine by the CPI was riven by rising foo, energy, an housing costs. In Germany, energy an housing prices were the primary sources of high CPI growth relative to the GDP eflator. In both Germany an the Unite States, investment goos prices grew much slower than the CPI. In the Unite States, prices for information technology goos, which represent a significant share of investment, have since the 1980s fallen rastically (e.g. the cost of a computer with 1 gigabyte of RAM) which reuce GDP eflator growth. Table 2 illustrates the relative importance of labour's terms of trae to each country's overall prouctivity-wage gap. The relative importance of labour's terms of trae in Norway is in part riven by it being the component largest of any of the 11 countries (Chart 5), but the relative size is even greater ue to Norway's relatively small overall gap. Similarly, labour's terms of trae make a larger absolute contribution to the gap in the Unite Kingom than in the Unite States or Germany because of the Unite Kingom's relatively small overall prouctivity-wage gap. Labour's share of income Labour's share of income measures the fraction of aggregate income in the economy (i.e. GDP) which is pai to workers as compensation for labour. Up until quite recently, labour's share of income was consiere constant by most economists, so much so that it became one of the main stylize facts presente in introuctory macroeconomics courses. Labour's falling share of income over the past ecaes in OECD countries has been well ocumente (OECD, 36 N U M B E R 32, S P R I N G 2017

13 Chart 7: Labour's Share of Income, Percentage Point Contribution to the Gap, (per year) Note: * , , , Source: Table ; International Labour Organization, 2015). Chart 7 presents the percentage point contribution to the wage-prouctivity gap mae by changes to labour's share of income over time. Notably, in three of OECD's countries, Spain, Denmark, an France, labour's share of income either hel steay or improve. Labour's share of income fell the most in Irelan, in large part as a result of capital's share increasing as foreign firms move their heaquarters there ue to favourable tax treatment. So far as the importance of labour's share of income to the overall prouctivity-wage gap, Table 2 presents the per cent contribution it mae. In five of the 11 OECD countries stuie (Finlan, Irelan, the Netherlans, Norway, an Spain), labour's share of income mae a contribution well in excess of 50 per cent. Ultimately, a ecline in labour's share of income over the perio as a whole inicates that labour's bargaining power has been falling relative to that of capital. In terms of our ecomposition, a ecline in labour's share of income over time leas to an increase in the overall gap. The causes of labour's eteriorating bargaining power are hotly ebate. One of the most trumpete causes is globalization. Proponents argue that capital is far more mobile than labour in an increasingly globalize worl, which makes the threat of outsourcing an offshoring far more creible. Due to the threat of offshoring from countries with less strict labour regulations an lower labour costs, workers are increasingly force to accept lower wages. Some argue that labour's eteriorating bargaining power is less a matter of globalization an more a matter of technological change which is biase against labour. For example, the OECD (2012) argues that the sprea of information an communication technologies have le to major innovation an prouctivity gains over recent ecaes, but have also ha the effect of replacing workers altogether. The result is an increase in capital's bargaining power, an a ecrease in labour's particularly for workers in highly repetitive jobs which naturally len I N T E R N A T I O N A L P R O D U C T I V I T Y MON I T O R 37

14 Chart 8: Hourly Real Wage Growth for Meian an the Top One Per Cent, Selecte OECD Countries, (average annual per cent change) e Note: * , , , Source: Househol Survey Microata from Luxembourg Income Survey D KW Chart 9: Ratio of the Average Wage of the Top One Percent to the Meian Wage, Selecte OECD Countries, 2013 e e Note: *last year available is Source: Househol Survey Microata from Luxembourg Income Survey themselves to automation. Structural an institutional reforms may also have contribute to the reuction of labour s bargaining power. Alternative Measures of Wage Inequality The measure of wage inequality use in the analysis so far has been to compare the national average meian to hourly real wages. While this measure captures whether or not the istribution is becoming more positively skewe overall, it oes not capture evelopments throughout the istribution. For example, it may be the case that the meian is growing at a similar rate as the mean, but the tails of the istribution are being stretche apart as those on the left tail experience little growth an those on the right tail experience extreme growth or vice versa (i.e. the istribution's skew may remain largely unchange but the height of the istribution may be changing). There are several alternative measures of wage inequality, such as the wage Gini coefficient, the ratio of the 90th percentile of wages to the 10th per- 38 N U M B E R 32, S P R I N G 2017

15 Table 4: Top One Percent's Share of Total Labour Income, per cent, Selecte OECD Countries, 1986 an 2013 eee W h^ e e e h< ee e e e ' e e & e e / e e E e E e & e e ^ Note: *last year available is Source: CSLS calculations base on microata from Luxembourg Income Survey centile, or the ratio of the 90th percentile to the 50th, as well as growth for the top one per cent of wage-earners. Unlike SNA ata, househol surveys allow us to investigate how the wage istribution is evolving by focusing on the wage growth experience by certain percentiles or subsamples. Mechanically, this ecomposition is the same as the ecomposition we have been employing throughout this article, with one change: we supplement the meian with a percentile such as the top 1 per cent of the statistic of interest. Chart 8, which is base on microata, compares the real hourly wage growth of the meian worker in a given country with the average real hourly wage growth of workers in the top 1 per cent of wage-earners. It shows the sobering fact that the wages of highly pai workers have greatly outpace the wages of workers in the mile of the wage istribution. In all countries except Spain. It is also important to consier the levels of wages to gauge the egree of wage inequality in these countries. Chart 9 provies the ratio of the wage of the top one percent to meian wage as a measure of the level of wage inequality in a given country. The Unite States has by far the highest level of wage inequality using this measure, with the top one percent earning on average more than 12 times meian income. Canaa an the Unite Kingom also have higher levels of inequality than the other 8 countries. The proportion of the wage income of the top one per cent in total labour income has grown (Table 4). The OECD (2012) has ocumente labour's falling share of income, an foun that removing the top one percent from labour income ouble the rate of ecline of labour's share of income in Canaa an the Unite States. In fact, the removal of the top one percent from total labour income hastene the ecline in labour's share of income in all of the OECD countries they stuie except Spain. I N T E R N A T I O N A L P R O D U C T I V I T Y MON I T O R 39

16 Table 5: Decomposition of the Gap between Labour Prouctivity an Real Wages Growth at Six Points in the Wage Distribution, in Selecte OECD Countries Unite States, Labour Prouctivity Per cent per year Hourly Real Wage Prouctivity- Wage Gap Inequality Percentage Point Contribution Employer Labour s Social Terms of Contribution Trae Labour s Share of Income e e e e D e e ee e e e e e e e KW e e e e e Z e e e e D e e e e e Canaa, e ee D e e e e e e ee KW e e e ee Z e e e D e ee Denmark, e e e ee e D e ee e ee e e e e e ee e KW e e e e ee e Z e e e e ee e D e e ee e Finlan, ee e D e e e e e KW ee e e Z e e D e e France, e e e e e e D e ee e e e e e e e e e KW e e ee e e Z e ee e e e D e e e e UK, e D e e e e e e KW e e Z e e D e Irelan, e e e D e e ee e e e e e KW e e e e Z e ee e D e e e 40 N U M B E R 32, S P R I N G 2017

17 Netherlans, Labour Prouctivity Per cent per year Hourly Real Wage Prouctivity- Wage Gap Inequality Percentage Point Contribution Employer Labour s Social Terms of Contribution Trae Labour s Share of Income e e e e D e ee e e e e e e e e e KW e e e e Z e e e D e e e e e Norway, e e e e e D e e e e e e e e e e e e e KW e e e e e e Z e e e e e D e e e e Spain, e e e D e e e e e e e e KW e e e e Z e e e D e e e Germany, e e ee e e D e e e e e e ee e e e KW e e e e e Z e e e e e D e e e e e Source: CSLS Calculations nase on microata from Luxembourg Income Survey We can also consier the first an thir quartiles (i.e. the 25th an 75th percentiles), as well as the prevailing wage of the top one percent, the rest, or those below meian wage. 11 For the three latter subsets, we use the average hourly real wage of the subset in our ecomposition. We use the average of the subset rather than the meian of the subset because we want to capture the effect of high or moest-income earners pulling the average in one irection or another: we want to estimate how wages have change for the group on the whole. Table 5 isplays the ecomposition results using alternative wage measures in place of the meian wage for all 11 countries in our ataset (The results using the meian are also isplaye for the sake of comparison.) The prouctivitywage growth gap in the Unite States is largest when the meian wage is use. This reflects the fact that real wage growth over the perio was lower at the meian than at other points throughout the wage istribution. That being sai, four of the five alternative real wage measures grew more slowly than labour prouctivity over the perio. Only the wages of the top one per cent grew faster than prouctivity growth. 11 The latter three groups are subsets of the population. The top one per cent the subset of all those with income above the 99th percentile. The Rest is the complement of the top one per cent, an consists of all those who o not earn an income above the 99th percentile. The below meian wage set is, as the name states, the subset of all of those with income below the 50th percentile. I N T E R N A T I O N A L P R O D U C T I V I T Y MON I T O R 41

18 Chart 10: Real Hourly Wage Growth by Percentile, Unite States, (average annual rate of change), W e e e e e e e e ee W Source: CSLS calculations base on microata from Luxembourg Income Survey Chart 10 provies a closer look at the iniviual percentiles for the Unite States. Hourly real wage growth in the Unite States for the perio stuie was largely below 0.40 per cent per year roughly between the 35th an 70th percentiles. Otherwise, hourly real wage growth tene to be far closer to or above average hourly real wage growth for the whole wage istribution (0.67 per cent per year). By focusing on the meian we inavertently chose the group in the Unite States which appears to haveexperience the least hourly real wage growth from 1986 to These results convey a narrative all too familiar. In the Unite States, the mile income earners have experience far less growth over the past ecaes than high or moest income earners. The same picture obtaine for most of the other 10 countries in Table 5. The hourly wage growth of the top one per cent excee growth of prouctivity in all countries, even in the three countries where wage growth ha exceee prouctivity growth. Conclusion Labour prouctivity growth outstrippe meian hourly real wage growth for the past few ecaes in eight of the 11 OECD countries stuie. For these countries, we ecompose the growing prouctivity-wage gap into four components: inequality, employement contributions to social insurance, labour's terms of trae, an labour's share of income. The size of the prouctivity-wage gap varie by country, as i the components riving its growth. Increasing inequality an labour's falling share of income increase the prouctivity-wage gap in most of the countries stuie. The prouctivity-wage gaps in the Unite States an Germany were significantly larger than any of the other countries stuie. The former's gap was largely riven by an labour's increasingly unfavourable terms of trae, while the latter's gap was riven by these two factors an a ecline in labour s share of income. We also show that espite inications of growing wage inequality in 10 of the 11 OECD countries, our inequality measure fails to capture a number of aspects of the overall evolution of the wage istribution. For example, while the ratio of average to meian wages in the Unite States has shown overall increases, there has been increase equality between mile an moest income earners. Future research shoul seek to reuce the likihoo that measurment error or efinitional ifferences across countries are responsible for ifferences in trens. Wage ata may be improve by using househol surveys irectly as oppose to accessing them through the Luxembourg Income Stuy. For example, using the 42 N U M B E R 32, S P R I N G 2017

19 Labour Force Survey for Canaa it is possible to create an annual wage series without neeing to interpolate missing values from 1997 to The lack of inclusive growth we observe in many OECD countries has significant societal implications. There may be less political support for prouctivity-enhancing policies in the future if the benefits of prouctivity growth are not share equitably. The incentives for employees to work har may iminish if they believe that they are not receiving their "fair share" of the firm's prouctivity gains. Finally, the current taxes an transfers system may not be well equippe to offset the growing tren of wage inequality among workers if it was esigne assuming labour prouctivity growth will lea to real wage growth for all workers. References Bartlett, Ranal, an Stephen Tapp (2012) An Assessment of Canaa's Labour Market Performance, Office of the Parliamentary Buget Officer, Available at: efault/files/files/files/ Labour_Note_2012_EN.pf. Bick, Alexaner, Nicola Fuchs-Schüneln, an Davi Lagakos (2016) "How o Average Hours Worke Vary with Development? Cross-Country Evience an Implications," NBER Working Paper Available at: w Felstein, Martin S. (2008) "Di Wages Reflect Growth in Prouctivity?" NBER Working Paper Available at: w13953.pf. Fisher, Tony an Doug Hostlan (2002) "The Long View: Labour Prouctivity, Labour Income, an Living Stanars in Canaa," Review of Economic Performance an Social Progress, pp (Ottawa: Center for the Stuy of Living Stanars an Montreal: Institute for Research on Public Policy). International Labour Organization (2015) Global Wage Report 2014/15: Wage an Income Inequality. Available at: wcmsp5/groups/public/---greports/---comm/ ---publ/ocuments/publication/ wcms_ pf. Mishel, Lawrence an Kar-Fai Gee (2012) "Why Aren't Workers Benefiting from Labour Prouctivity Growth in the Unite States?" International Prouctivity Monitor, No. 23. Spring, pp OECD (2014) All On Boar: Making Inclusive Growth Happen. Available at: inclusive-growth/all-on-boar-making-inclusive-growth-happen.pf. OECD (2012) "Labour Losing to Capital: What Explains the Declining Labour Share?" OECD Employment Outlook. Available at: EMO%202012%20Eng_Chapter%203.pf. Pessoa, Joao Paulo, an John Van Reenan (2012) "Decoupling of Wage Growth an Prouctivity Growth? Myth an Reality," Resolution Founation. Available at: Decoupling-of-wages-an-prouctivity.pf. Schwellnus C., A. Kappeler an P. Pionnier (2017) The Decoupling of Meian Wages from Prouctivity in OECD Countries, International Prouctivity Monitor, Vol. 32, pp Sharpe, Anrew, Jean-François Arsenault, an Peter Harrison (2008a) "Why Have Real Wages Lagge Labour Prouctivity Growth in Canaa?" International Prouctivity Monitor, No. 17. Spring, pp Sharpe, Anrew, Jean-François Arsenault, an Peter Harrison (2008b) "The Relationship between Real Wage Growth in Canaa an OECD Countries," CSLS Research Report Number 8, Schwellnus C., A. Kappeler an P. Pionnier (2017) The Decoupling of Meian Wages from Prouctivity in OECD Countries, International Prouctivity Monitor, No. 32. Spring, pp Uguccioni, James, Alexaner Murray an Anrew Sharpe (2016) "Decomposing: Labour Prouctivity an Real Wage Earnings in Canaa, ," CSLS Research Report Number15, November.. Uguccioni, James an Anrew Sharpe (2016) "Decomposing the Prouctivity-Wage Nexus in Selecte OECD Countries, ," CSLS Research Report Number 16, November.. I N T E R N A T I O N A L P R O D U C T I V I T Y MON I T O R 43