Selected Issues

Abstract

Selected Issues

Understanding Wage Growth in Japan1

Lackluster nominal wage growth has frequently been highlighted as a key reason for Japan’s continued struggle to reflate the economy. This chapter takes a closer look at why wage inflation has been missing in Japan. The empirical results suggest that while cyclical factors (e.g., labor market slack and inflation expectations) play an important role in driving wage growth, the main reason for the “missing wage inflation” is the downward pressure exerted by structural factors (e.g., dual labor market effects and wage rigidities).

1. Wage inflation in Japan has been largely absent since the mid-1990s. Annual wage growth fell sharply following the bursting of the asset bubble in the early 1990s. By the end of the decade it had dropped below zero (see figure). The lackluster wage growth continued between 2000 and 2007, and remained in negative territory during the initial phase of the Global Financial Crisis (GFC). In the post-crisis era, however, wage growth has been on a slight upward trend. Nevertheless, wage growth remains well below the government’s three percent goal, and continues to pose a major roadblock to the Bank of Japan’s (BoJ) efforts to reflate the economy. The low wage growth is particularly puzzling given the sharp drop in unemployment in recent years, triggering renewed interest in finding the causes for this “missing wage inflation” (e.g., Iwasaki et al, 2018; Bank of Japan, 2017).

uA05fig01

Japan: Wage Growth and Unemployment Gap, 1992–20181

(In percent)

Citation: IMF Staff Country Reports 2018, 334; 10.5089/9781484386804.002.A005

1 The unemployment gap equals the demeaned unemployment rate.Sources: Haver and IMF Staff calculations.

2. This chapter attempts to shed light on why wage inflation has been missing in Japan, distinguishing between cyclical and structural drivers. By estimating a set of wage Phillips curves (WPCs)—augmented with structural factors—several interesting results emerge. First, unemployment and inflation remain key drivers of wage growth. Second, the response of wage growth to labor market slack exhibits downward rigidity, causing the slope of the WPC to flatten at higher rates of unemployment. Third, the pass-through from productivity gains to nominal wages is economically significant. Fourth, labor market duality tends to negatively affect wage growth through its “composition effects,” but more importantly by flattening the WPC. Finally, distinguishing between cyclical and structural factors, the absence of wage inflation in recent years, despite labor market tightening, is due primarily to structural factors—the lack of price pressure on consumer goods has meant less upward pressure on wages.

A. Cyclical and Structural Drivers of Wage Growth in Japan

3. Nominal wages appear responsive to economic fluctuations, but the relationship changes over the business cycle and over time.

  • Labor market slack. In economic downturns, demand for goods and services falls, resulting in downward pressure on both prices and output. To maintain profitability, Japanese firms seek to constrain labor costs by constraining employment and wage growth. One would thus expect wage growth to be pro-cyclical and negatively correlated with unemployment. Indeed, this relationship broadly holds in Japan over the period 1993–2018 (see figure). However, the relationship also appears to change over time. It was particularly strong in the 1990s, weakened significantly between 2000 and 2007, but reappeared after 2011.

  • Inflation. When the cost of living increases, workers will demand higher nominal wages to maintain the purchasing power of their earnings. On the other hand, if output prices fall, firms will push for lower nominal wages to maintain profit margins. Moreover, given wage rigidities, expected inflation is also likely to be an important determinant of current wages. Not surprisingly, wage growth and inflation do indeed exhibit a strong positive relationship (see figure).

  • Wage rigidities. Wage growth may also vary depending on whether the economy is in a weakening or strengthening phase. For instance, in a cross-country analysis, IMF (2017) finds that, at any given level of labor market slack, wages tend to adjust faster when the economy is entering a recession than when it is exiting a recession. Of course, the opposite could also be true. That is, due to rigidities, wage growth may be slow to adjust in the initial stage of the recession only to pick up once the recovery is well underway. In addition, a large body of literature has argued that firms may find it harder to lower nominal wages in downturns. This would suggest an L-shaped relationship between wage growth and unemployment. In fact, the reasons why wage growth seemed unresponsive to unemployment during the period 2000–07 may be that unemployment was above its long-term level for most of this period and wages were unable to adjust downward.

uA05fig02

Japan: Annual Wage Growth and Unemployment Gap, 1993–20181/

(In percent)

Citation: IMF Staff Country Reports 2018, 334; 10.5089/9781484386804.002.A005

1/ The unemployment gap equals the demeaned unemployment rate.Sources: Haver Analytics and IMF staff calculations.
uA05fig03

Japan: Annual Wage Growth and Inflation, 1993–2018

(In percent)

Citation: IMF Staff Country Reports 2018, 334; 10.5089/9781484386804.002.A005

Sources: Haver Analytics and IMF staff calculations.

4. Trend productivity growth has likely also influenced wage growth dynamics. Holding wages and prices constant, the more productive workers are, the lower is the cost of producing one additional unit of output. The consequent increases in profit per unit of output, encourages firms to expand production and hire more workers. Eventually, wages will be bid up and the cost per unit of output rises to reflect the higher level of labor productivity. Hence, we would expect nominal wage growth and trend labor productivity growth to be positively correlated. This prediction appears supported by the unconditional correlation between the two variables (see figure).

uA05fig04

Japan: Annual Wage and Trend Productivity Growth, 1993–20181/

(In percent)

Citation: IMF Staff Country Reports 2018, 334; 10.5089/9781484386804.002.A005

1/ Trend productivity is measured as HP-filtered real output per hour.Sources: Haver Analytics and IMF Staff calculations.

5. The impact of labor market duality on wage growth could work through several channels, but the net effect may be ambiguous. Employment in Japan can generally be divided into regular or non-regular workers. Workers employed under regular contracts enjoy higher wages and job security than non-regular workers. Non-regular workers are predominantly employed part-time and are not unionized. Helped by the passing of major labor market reforms in the late 1990s and early 2000s, the share of non-regular workers has gradually increased in Japan (see figure), reflecting a desire by firms to lower overall wage costs and increase employment flexibility.2 The increased labor market duality is likely to have affected wage growth through several channels.

uA05fig05

Japan: Labor Market Duality, 1984–2018

Citation: IMF Staff Country Reports 2018, 334; 10.5089/9781484386804.002.A005

Source: Ministry of Internal Affairs and Communications.
  • Involuntary part-time employment effect. Because some fraction of part-time workers would prefer to be employed full-time, a high share of part-time workers may reflect a certain degree of labor market slack not captured by the overall unemployment rate. Since most non-regular workers are employed part-time, a rise in the share of non-regular workers would be indicative of growing labor market slack.

  • Composition effect. With non-regular workers on average being paid less, a higher share of non-regulars would automatically translate into a lower overall wage level. Hence, wage growth should be negatively related to the change in the share of non-regular workers.

  • Inside-outside effect. How wages respond to labor market slack may differ between regular (protected insiders) and non-regular workers (unprotected outsiders). For instance, Bank of Japan (2017) argues that wages of regular workers may not respond much to labor market slack as labor unions (which overwhelmingly represent regular workers) prioritize job security over wage increases even under tight labor market conditions. Hence, an increase in the share of non-regular workers should lead to a steeper WPC.

  • Labor supply elasticity effect. Because the supply of non-regular workers responds with relatively higher elasticity to wage increases (e.g., labor force participation rates of women and older workers have increased in tandem with demand for non-regulars), the rising share of non-regular workers may have increased the overall labor supply elasticity and hence caused the slope of the WPC to flatten.3, 4

B. Estimating Wage Phillips Curves Augmented with Structural Drivers

6. Several WPC specifications are estimated to assess the relative importance of cyclical and structural drivers of wage growth in Japan. The methodology follows IMF (2017), but is extended to incorporate downward wage rigidity and dual labor market effects. Table 1 displays the results from five different WPC specifications.

  • Baseline specification (column 1). In the baseline regression, annual wage growth—measured by scheduled cash earnings per hour—is regressed on (i) lagged annual CPI inflation, (ii) expected CPI inflation five years ahead based on consensus survey data, and (iii) the unemployment gap, defined as the demeaned unemployment rate. Two dummies for the 1997 and 2014 VAT rate increases are also included.5

  • Wage rigidity specification (column 2). To test for wage rigidities (as discussed above), the baseline regression is augmented in the following manner. First, the change in the unemployment gap is included to capture any potential lagged response of wage growth to labor market slack. Second, an interaction term between the squared unemployment gap and a binary dummy variable—taking the value one if unemployment is above its long-run average—is added to allow for non-linearities in the WPC and downward wage rigidity.

  • Trend productivity growth and labor duality (column 3–5). HP-filtered real output per hour is included to as a measure for annual trend productivity growth. Labor market duality is modeled in three ways. First, the ratio of part-time workers is included to capture added slack from involuntary part-time employment. Second, the change in the ratio of part-time workers is used as a proxy to account for the composition effect from a rising share of non- regular workers. Third, the ratio of part-time workers is interacted with the unemployment gap to investigate how labor market duality affects the slope of the WPC.

Table 1.

Japan: Estimates of Wage Phillips Curves Augmented with Productivity, Wage Rigidity, and Labor Market Duality

article image
Source: IMF staff calculations.Note. Standard errors are made robust to autocorrelation and heteroscedasticity. Standard errors in parentheses. * p < .10; ** p < .05.

Wage growth is measured as annual average scheduled cash earnings per hour

Lagged inflation is the year-on-year CPI inflation lagged one period. Inflation expectations is the 5-year consensus forecast.

The unemployment gap is the demeaned unemployment series, and the sign dummy takes the value one when the unemployment gap is negative.

Trend productivity growth is HP-filtered real output per hour.

7. Overall the empirical results suggest that both cyclical and structural factors play an important role in driving wage growth in Japan.

  • Labor market slack. Wage growth is negatively related to the unemployment rate across specification, indicating that wage growth does respond to the business cycle. However, the share of part-time workers does not seem to affect directly wage growth to a significant degree, suggesting that the dual labor market does not contain much “hidden” slack.

  • Inflation. A one percent increase in lagged inflation translates into a 0.3–0.4 percentage point increase in wage growth. The impact of inflation expectations on wage growth is large and significant for most of the specifications, but its economic importance falls as structural features are added to the model. The average impact (across specification) of expected inflation on wage growth is about 0.8. Taking these estimates at face value, if BoJ would successfully anchor actual and expected inflation at its 2 percent inflation target, wage growth would on average—abstracting from productivity gains—equal about 2 percent over the business cycle.

  • Wage rigidities. The change in the unemployment rate has a positive and significant coefficient. This is contrary to the findings in IMF (2017) and would indicate that, given the degree of labor slack, wages tend to adjust slower when the economy is weakening than when it is strengthening. Hence, wage growth appears to react with a lag to labor market slack. The positive and significant coefficient on the interaction term between the squared unemployment gap and the sign dummy indicates that downward rigidity is an important factor and that the WPC is indeed L-shaped in Japan.

  • Productivity. Trend productivity growth enters the regression positively and is statistically significant, suggesting that productivity gains are an important driver of wage growth in Japan. The point estimate (in column 5), indicates that over two thirds of the productivity gain is passed on to wages.

  • Labor market duality. The interaction terms between the unemployment gap and the share of part-time employment share is statistically significant and positive. This indicates that an increase in the part-time employment share causes the slope of the WPC to decline (as predicted by the supply elasticity effect). In fact, the rising share of part-time workers has caused the slope of the WPC to fall from around 1.4 percent in 1993 to 0.6 percent in 2018, suggesting that the impact on wage growth from a fall in the unemployment rate in 1993 would have been twice as large as today. In addition, as expected, the coefficient on the change in the share of part-time workers enters negatively and is significant, showing that the level composition effect has indeed constituted an important drag on wage growth.

uA05fig06

Japan: Slope of Wage Phillips Curve, 1993–2018

Citation: IMF Staff Country Reports 2018, 334; 10.5089/9781484386804.002.A005

Source: IMF staff calculations.

8. Japan’s missing wage inflation is due primarily to dual labor market effects and low consumer price inflation. Using the WPC estimation from column 5 in Table 1, it is possible to compare various drivers of wage growth in recent years to those in the pre-GFC period 2000–07 (see figure). For instance, in 2018, wage growth was about one percent higher than the average wage growth in the pre-GFC period. Abstracting from labor market duality and wage rigidities, cyclical factors (unemployment and inflation) and productivity gains should have generated wage growth about 4 percent higher than in the pre-GFC period. However, wage rigidities and dual labor market effects shaved of about 3 percent from wage growth. The flattening of the WPC accounted for about two thirds of the drag and wage rigidities for about one third.

uA05fig07

Drivers of Annual Wage Growth in Japan, 1994–2017

(In percent, relative to 2000–07 average)

Citation: IMF Staff Country Reports 2018, 334; 10.5089/9781484386804.002.A005

Source: IMF Staff calculations.

C. Conclusion

9. The empirical results presented in this chapter indicate that both structural and cyclical factors are important drivers of wage growth, although they tend to work in opposite directions. Despite reduced labor market slack in recent years, wage inflation has only showed a slight pick-up. Estimation of the wage Phillips curve (WPC) augmented with structural drivers, shows that the main reason for this “missing wage inflation” is the downward pressure exerted by dual labor market effects and wage rigidities. The flattening of the WPC due to the large share of non-regular workers has been a particularly important drag on wage growth. However, looking forward, as the pool of untapped non-regular workers wanes and their labor supply elasticity rises, the WPC may steepen.

10. The results also highlight the importance of implementing a comprehensive and coordinated policy package to kick start wage-price dynamics and address labor market duality. Reducing the negative effects from labor market duality could have a significant impact on wage growth both by reducing the wage gap between regular and non-regular workers (i.e., reversing the composition effect), but also by making demand stimulus more effective through a steepening of the WPC. Higher wage growth would in turn help stimulate demand for goods and services and lead to higher inflation, which would feed back into higher wage growth. Income policies to boost wages could also provide an initial boost to wage growth, but have to be backed by significant demand support to be successful.

References

  • Bank of Japan, 2017, “Features of Japan’s Labor Market and Macro-Level Wages,” Box 2, Outlook for Economic Activity and Prices, July, Bank of Japan: Tokyo.

    • Search Google Scholar
    • Export Citation
  • Bank of Japan, 2018, “The Recent Increase in Labor Supply and Wage Developments,” Box 1, Outlook for Economic Activity and Prices, July, Bank of Japan: Tokyo.

    • Search Google Scholar
    • Export Citation
  • International Monetary Fund, 2017, “Recent Wage Dynamics in Advanced Economics: Drivers and Implications,” Chapter 2, World Economic Outlook, October, Washington D.C.

    • Search Google Scholar
    • Export Citation
  • Iwasaki Y., I. Muto, and M. Shintani, 2018, “Missing Wage Inflation? Estimating the Natural Rate of Unemployment in a Nonlinear DSGE model.” 2018 BOJ-IMES Conference, IMES, Bank of Japan.

    • Search Google Scholar
    • Export Citation
  • Gali, J., 2011, “The Return of the Wage Phillips Curve,” Journal of the European Economic Association, Vol. 9, pp. 43661.

  • Watanabe, H. R., 2018, “Labour Market Dualism and Diversification in Japan,” British Journal of Industrial Relations, Vol. 53, Issue 3, pp. 579602.

    • Search Google Scholar
    • Export Citation
1

Prepared by Niklas Westelius (APD).

2

Watanabe (2017) suggest that the motivation for the labor market reforms in the late 1990s and early 2000s reflected the need by Japanese companies to increase functional flexibility and decrease production costs in response to intensified regional economic competition.

3

Bank of Japan (2018) finds that the wage elasticity of labor supply of female and senior part-time workers is higher than working aged men, and argues that this has reduced the impact of increased labor demand on wages.

4

See Gali (2011) for a theoretical derivation of the negative relationship between labor supply elasticity and the slope of the WPC.

5

Dummies for VAT rate increases in 1997 and 2014 are also included in all specifications. Results indicates that while the 1997 VAT rate rise did affect wages positively, the 2014 VAT rate increase did the opposite.

Japan: Selected Issues
Author: International Monetary Fund. Asia and Pacific Dept