A Study on the Downward Nominal Wage Rigidity. in Korea

Transcription

1 A Study on the Downward Nominal Wage Rigidity in Korea Mihye Lee Su-mi Na October 12, 2014 Abstract This paper examines the downward nominal wage rigidity in Korea. It explores the wage rigidity using aggregate and individual level data. The results show that, on average, wages have been flexible. However, micro-data suggests that nominal wages are downwardly rigid most of the time. We document the evidence based on the average wage changes of all workers and of individual workers. In addition, we show that the downward nominal wage rigidity can be different across industries at industry and individual levels in each industry: The downward nominal wage rigidity is higher in the service sector than in the manufacturing sector at the individual level even though, the manufacturing sector exhibits smaller volatility in its wage growth rate than the service sector. Corresponding author, Economist, Economic Research Institute, Bank of Korea (Tel: , Department of Economics, Seoul National University( 1

2 1 Introduction Since the Global Financial Crisis in 2008, many countries have experienced a sharp rise in unemployment as they entered an economic recession. After several years of financial crisis, an unemployment did not go back to its previous level; rather it exhibited sustained high unemployment levels for a long time. This is somehow puzzling. One might think that when the unemployment rate is high then wages (the price of labor) would go down. In such a situation, unemployment would be offset somehow through these wage cuts. In reality, however, this adjustment barely happens. Schmitt-Grohé and Uribe (2013) show that countries in the Eurozone failed to adjust wages (measured as nominal hourly wages) and this has resulted in widespread unemployment. The phenomenon that (nominal) wage is resistant to change is called downward nominal wage rigidity (DNWR) 1. In particular, the downward nominal wage rigidity can be costly in a low inflation environment. Many countries have faced this situation in recent years 2. For example, firms can adjust their labor cost by paying relatively low real wages without cutting their nominal wages when inflation is high (they can offset the increase in labor cost with high inflation so that the real labor cost will not rise as much as the increase in labor cost). In contrast, it becomes harder for firms to avoid large labor cost in a low inflation environment, since firms cannot cut their real labor cost without actual wage cut different from high inflation environment. In such a case, the unemployment rate may rise since the only way for a firm to reduce its labor cost is through an adjustment in employment. It is noteworthy, therefore, the existence of the downward nominal wage rigidity may be an important force in accounting for high unemployment with low inflation rates. Hence, examining the existence or degree of the downward nominal 1 There are several reasons why firms do not cut their employees wages. One of the possible explanations is that firms fear moral hazard possibly triggered by wage cuts or workers resistance. For further explanations, please see Akerlof et al. (1996) 2 The U.S. inflation rate is below 2% as of September The same holds for Korea also. 2

3 wage rigidity may provide some insight into how the labor market responds flexibly to a sudden (exogenous) negative shock or to changing economic conditions. The large amount of existing literature has examined micro data on the distribution of individual wage growth. In addition, many scholars have attempted to explain sustained high unemployment rates in the U.S. and Europe by looking at the existence of nominal or real wage rigidity since the Great Financial Crisis. Some of those studies show that the existence of nominal wage rigidity might have contributed to sustained high unemployment rates after the recent financial crisis. Similarly, studies unveil some factors that are responsible for high unemployment rates in Japan during its lost decade with concern for deflation. Nevertheless, there has be a little research on the nominal wage rigidity in Korea. Our paper attempts to provide some evidence of nominal wage rigidity using micro and macro data. In the past, the lack of a proper individual wage data hinder researchers from examining this issue. The household income panel data has been available since 1999 so this enables us to document some evidence based on the existing literature and provide some evidence to support the existence of the nominal wage rigidity in Korea. Related Literature Several strands of literature have studied the nominal wage rigidity. Before proceeding, we present some literature that are closely linked to this paper. A strand of literature assesses the extent of wage rigidity based in micro data by looking at the distribution of wage growth (McLaughlin, 1994, 1999, 2000; Lebow et al., 2003; Kuroda and Yamamoto, 2003). Some of those papers try to find whether the degree of wage rigidity is related to inflation rates. Daly et al. (2012) show that real wage growth in the U.S. did not adjust downward with low inflation during the recent recession that followed the financial crisis in They document that the 3

4 fraction of workers with wage freezes (equivalently, no wage change) is relatively large, which supports the existence of nominal wage rigidity in the U.S. Akerlof et al. (1996) also argue that it is hard for firms to adjust labor costs with low inflation which leads distortions into the labor market. Similar to Akerlof et al. (1996), Bauer et al. (2003) study the relationship between inflation and nominal real wage rigidity in Germany. In contrast to the previous findings, there is a positive relationship between the extent of real wage rigidity and inflation, and the nominal wage rigidity falls with inflation, as opposed in the case of real wage rigidity, thus confirming the findings from Akerlof et al. (1996). Those studies show that nominal wage rigidity can be more severe in a low inflation environment. It is also noteworthy that there have been relevant studies on whether there exists the nominal wage rigidity in Japan. For example, using industry and aggregate level data Kuroda and Yamamoto (2014) and Kimura and Ueda (2001) show that the nominal wage became downwardly rigid before the recession but, was flexible afterwards in response to the economic conditions (Kuroda and Yamamoto, 2005). However, Yasui and Takenaka (2005) deliver different results from the previous studies; they find that nominal wage rigidity seemed to be present. Most existing studies are quite limited in providing comparable results for different countries subject to the data availabe. Recently, Dickens et al. (2007) present some evidence which helps us compare the extent of wage rigidity across different countries. They show that the degree of wage rigidity is significantly different across countries and time periods. Schmitt-Grohé and Uribe (2013) also provide some evidence based on Eurozone countries and argue that it might have contributed to sustained high unemployment rates in the region after the financial crisis. Nevertheless, those studies do not include Korea, so it is impossible to compare the wage flexibility of Korea with other countries. 4

5 Very few studies of downward wage rigidity exists in Korea, especially in the 2000s. Lee (1999) and Yoon (1994) investigated this issue but it is hard to find any literature. Using Daewoo Panel from 1992 to 1995, Lee (1999) documents that real wage rigidity is acyclical. Yoon (1994) assesses a wage rigidity based on the wage equation using aggregate level data. This paper tries to provide some empirical evidence of the existence of nominal wage rigidity using micro and macro level data. As well we look for possible implications on the labor market. This paper is organized as follows. Section 2 explains how we measure wage rigidity. Section 3 tries to compare the degree of nominal wage rigidity across countries using the relationship between inflation and degree of nominal wage rigidity and the relationship between growth rate and degree of nominal wage rigidity. Section 4 presents some evidence of how wages have evolved based on average wage growth rates. Section 5 presents the degree of nominal wage rigidity using individual wage data. Section 6 examines the nominal wage rigidity based on industry level. Section 7 concludes and suggests some related issues for future research. 2 Measuring the Downward Nominal Wage Rigidity There are several ways to assess the downward nominal wage rigidity depending on the data that are used and how we define this rigidity. Broadly speaking, the approaches can be classified into two categories based on the data used as follows: macro level data that keeps tracks of average wage growth of all workers in the economy, and individual level data that captures wage changes in the sample. In this paper, we scrutinize the degree of nominal wage rigidity based on the av- 5

6 erage wage growth (or change) of all workers, along with individual-level data 3. The findings from those two different sources of data reveal that wages on average have been flexible. Average wages changed during the late 1990s and in the recent financial crisis. However, the evidence based on micro-data suggests that nominal wages are downwardly rigid most times. In addition, we show that the downward nominal wage rigidity can be different across industries at the industry- and individual-level in each industry. As mentioned, we provide descriptive evidence of downward nominal wage rigidity based on the two data sets. First, tusing individual-level data, we calculate the wage growth of workers who stayed their jobs at least for two consecutive years and who reported their earnings. Previous studies draw the distribution of wage changes and assess the existence of the downward of nominal wage rigidity : i) whether there is a spike at zero ii) when the distribution is skewed to the right. Instead of focusing on the distribution of individual worker s wage growth, Dickens et al. (2007) suggest alternative measures of the extent of downward nominal and downward real wage rigidity. According to Dickens et al. (2007) For downward nominal wage rigidity, our measure is straightforward. We assume that everyone who had a nominal wage freeze would have had a nominal wage cut in the absence of downward nominal wage rigidity n = f n f n + c n (1) denotes the fraction of workers who experienced nominal wage freezes and c n denotes the fraction of workers whose wages were cut. Thus, the nominal wage rigidity is expected to be high if the fraction of workers with wage freezes is large. 3 We use data from the Korea Labor and Income Panel Study (KLIPS) from 1999 to KLIPS is a longitudinal survey for households residing in urban area. The sample is about 5,000 households and it collects information on labor force supply and mobility, income, consumption, education, job training and etc. 6

7 Second, we evaluate the existence of downward nominal wage rigidity along with other macroeconomic variables. We show how the wage growth of all workers has evolved over time. For example, Schmitt-Grohé and Uribe (2013) point out that the existence of downward nominal wage rigidity in the Eurozone area played a significant role in accounting for hikes in unemployment rates since This approach enables us to examine the consequences of the downward nominal wage rigidity for macroeconomic adjustment to changes in unemployment when an economy is hit by a sudden shock. It is noteworthy that there can be discrepancies between the downward nominal wage rigidities observed in individual-level data and at the aggregate level (defined by the average wage growth of all workers in the economy) due to the following reason. Consider when firms are able to adjust their wage bills based on changes in the composition of workers. Even if there exists the downward nominal wage rigidity in an economy, it is hard to observe the downward nominal wage rigidity overall (Fuhrer et al., 2012). We provide some insight into whether this phenomenon holds in the case of Korea when we examine the existence of downward nominal wages at both individual and aggregate levels. 3 Cross-Country Comparison This section tries to compare the degrees of Korea s downward nominal wage rigidity with other countries. Most existing literature typically restricts their attention to an individual country subject to the availability of the data. However, Dickens et al. (2007) collect the data from 16 countries 4 and measure the overall downward nominal wage rigidity. We follow this approach and measure Korea s downward nominal wage 4 Ireland, Denmark, France, Belgium, UK, Switzerland, Austria, Germany, Italy, Netherlands, Finland, Norway, Greece, Sweden, US. Portugal 7

8 rigidity and then compare it with other countries. Some issues raise when we compare the nominal wage rigidity across countries. One of them might be the challenge of defining the nominal wage rigidity consistently. We use the measure presented in equation (1), since it is relatively easy to calculate and intuitive to understand. Secondly, we address the time period. When it comes to using the individual level data for cross-country comparison, it is hard to exactly match the time period for different countries. Dickens et al. (2007) simply use the average value of n for each country to resolve this issue, and we follow the same approach. So we calculate the n for each country and every year, if necessary, then calculate the average of these. The n which is the average value for different time periods from different countries is not ideal for comparing its absolute value, directly. Rather, we do the cross-country comparison in a way that corresponds to inflation and growth rate. Figure 1 plots the degree of the downward nominal wage rigidity 5 and inflation rates for each country. To find out whether countries with higher inflation will experience lower nominal wage rigidity, we run a regression with the measure of nominal wage rigidity n on inflation rates. The regression result shows that the slope is 0.04 which tells us that countries experiencing high inflation may have higher downward nominal wage rigidity. Most countries are around the regression trend line: Ireland has the least downward nominal wage rigidity while Portugal has the highest value. Countries with inflation rates of 2% 3% (Denmark, Germany and so on.) have the average n between 0.2 and 0.4. Korea also experienced a 2% 3% inflation rate from 1999 to 2010; however, it has a relatively higher degree of downward nominal wage rigidity compared to those countries with similar inflation rates. This indicates that Korea has a higher degree of downward nominal wage rigidity, given the inflation rate during this time. 5 It uses the average value of n for each country and time period. 8

9 Portugal 0.5 Degree of DNWR Denmark Netherlands UK Switzerland Korea Finland Norway Greece Germany Italy Austria US Sweden 0.1 Japan Belgium France Ireland Inflation rate(%) Figure 1: Inflation Rate and Downward Nominal Wage Rigidity For another cross-country comparison, we examine the relationship between real economic growth rates and the degree of downward nominal wage rigidity. Figure 2 shows a scatter plot of economic growth rates and the degree of downward nominal wage rigidity. Again, we run a regression using the degree of downward nominal wage rigidity as a dependent variable and real growth rate as an independent variable. Similar to the previous result, we obtain the slope 0.07 from this regression showing that countries with higher growth rates may undergo a higher degree of downward nominal wage rigidity. However, this relationship appears clearly within countries with growth rates below 3.5% and it is hard to observe this relationship in countries with higher growth rates. Figure 3 plots the times series of the nominal wage rigidity for Korea, Japan, U.S., U.K. and Switzerland for available periods. The data for Korea is drawn as in Figure 1 and Figure 2 and the data for other countries are from Kuroda and Yamamoto (2013), 9

10 Portugal Degree of DNWR US Sweden Denmark Greece Norway Italy Finland UK Netherlands Germany Austria Switzerland Korea 0.15 Japan Belgium France Growth rate(%) Figure 2: Real Economic Growth Rate and and Downward Nominal Wage Rigidity Elsby (2009), Smith (2000) and Fehr and Goette (2005). Overall, n does not exceed 0.3 for most countries. In contrast, the value is greater than 0.3 except in 1999 and it tends to increase over time. This simple comparison tells us that the nominal wage rigidity in Korea is greater compared to other countries. With the average degree of downward nominal wage rigidity, the data shows that countries with higher inflation experience higher downward nominal wage rigidity. Compared to other countries, Korea has a higher degree of downward nominal wage rigidity. We also find that there may be a positive relationship between growth rates and the degree of downward nominal wage rigidity; although the data exhibits a weak relationship with some outliers. Needless to say, we do not try to claim that higher inflation rates lead to higher downward nominal wage rigidity or that higher growth rates may cause higher downward nominal wage rigidity. Our purpose is to figure out the degree of the downward nominal wage rigidity of Korea, along with other countries. 10

11 Korea Japan US UK Switzerland Figure 3: Trend in Downward Nominal Wage Rigidity Based on Figure 1 and Figure 2, Korea seems to have experienced relatively higher downward nominal wage rigidity compared to countries with similar inflation or growth rates. 4 Downward Nominal Wage Rigidity in Korea So far, our analyses are mainly focused on the average degree of Korea s downward nominal wage rigidity compared to other countries. We have demonstrated that Korea has relatively higher downward nominal wage rigidity compared to countries with similar inflation rate growth rates. Then, what does this mean? If there exists the downward nominal wage rigidity, then it is natural to ask if it plays an important role in economic performance. In particular, does a higher degree of downward nominal wage rigidity cause higher unemployment when an economy is hit by a negative shock? To answer this question, we analyze overall wage growth rates along with other economic 11

12 variables such as inflation rate, growth rate and unemployment. This will give us some idea how the wage growth rate changes according to economic conditions in the case of Korea. Figure 4 plots Korean unemployment rates (%), average wage changes of all workers (%), and real economic growth rates (%) from 1990 to The economy was hit by negative shocks in 1997 and 2008, respectively. The growth rate plummeted to -5.7% in 1998 due to the currency crisis that happened in 1997 and it was near zero in 2009 after the global financial crisis. What is interesting here is that the nominal wage changes occur typically after a year. And the volatility of wage changes are significantly larger than expected based on the findings in the previous section. So it seems like wage changes are quite flexible in response to negative shocks. When we look at the unemployment rate, for example in 1998, it surged to 7.7% then decreased to around 3%. Overall, the unemployment returned to the average rate with an adjustment in wage changes. Figure 4 shows how wage growth and unemployment have evolved over time. To further examine the relationship between wage growth and unemployment, we now turn our attention to the wage Phillips curve. Figure 4 plots the wage growth rate and unemployment rate from 1987 to In the figure, we split the sample period into 5 sub-periods: i) (pre-asia currency crisis) ii) (during the crisis) iii) (before Great Financial Crisis) iv) 2009 v) (post-crisis). From 1987 to 1997, the wage Phillips curve appears to have the expected relationship between wage growth and unemployment. Unemployment ranges from 2% to 4% and when unemployment rises the wage growth rate falls. When the economy is hit by the crisis, the wage growth rate falls to almost 0% and the unemployment rate hikes to 7.7%. In the following year, the wage growth falls. If the wage growth rate did not fall, it is possible Korea may have undergone higher unemployment rates for a longer 12

13 Figure 4: Unemployment Rate, Nominal Wage Change and Economic Growth Rate ( ) 20 Unemployment rate(%) Nominal wage change(%) Real GDP growth rate(%) Source: Ministry of Employment and Labor, Statistics of Korea, Bank of Korea time. Besides unemployment, nominal wage changes are related to how firms determine prices. Under stable mark-up rules, Kuroda and Yamamoto (2013) show that inflation ( P/P ) can be written as follows: P/P = α(g w G l ) + (1 α)(e + p r G r ) (2) where G w is the nominal wage growth per worker, G l is labor productivity growth, (G w G l ) is the change in unit labor cost, e the exchange rate, p r is the change in the international price of raw materials G r is the productivity growth of a unit of raw materials, and α is the labor share, respectively. According to equation (2), as the labor share rises, nominal wage growth leads to higher inflation rates. Now, we try to 13

14 Figure 5: Korea Wage Phillips Curve: Nominal wage change rate(%) Unemployment rate(%) Source: Ministry of Employment and Labor, Statistics of Korea, Bank of Korea find how wage growth contributed to the changes in price levels. Figure 6 shows the nominal wage, labor productivity, price and unit labor cost (denoting 1995=100). Interestingly, nominal wage, productivity, price level and unit labor costs tend to rise over time. As shown, the increase in wage and labor productivity have different impacts on determining unit labor cost. For example, when nominal wage increases it results in a higher unit labor cost. In contrast, when labor productivity rises, it drives down the unit labor cost. The data shows that nominal wage increases at a much higher rate than labor productivity; therefore the unit labor cost increases slowly. In 1999, we observe the decline in the unit labor cost due to a decrease in nominal wages resulting from the crisis. Unlike in 1999, there was positive wage growth in 2010, however, the labor productivity grow more than the wage growth; thus the unit labor cost goes down in If we examine Figure 6 closely, the price level (CPI) grows rapidly during the period 14

15 Figure 6: Nominal Wage, Labor Productivity, Price and Unit Labor Cost: Nominal wage Productivity CPI Unit labor cost 220 index (1995=100) Source: Ministry of Employment and Labor, Statistics of Korea, Bank of Korea when the unit labor cost grows sluggishly. This tells us that the labor share (α) is not large enough 6. The findings in this section can be summarized as follows: First, the nominal wage growth adjusts when there are lags in economic conditions and the volatility of wage growth is larger than expected. The negative wage growth resulting from the negative shocks in 1999 and 2009 may have contributed to a stable unemployment rate. The data shows that nominal wages grew rapidly but their impact on price levels is somehow offset due to the increase in labor productivity and the small fraction of the labor share. This suggests that other factors may have contributed to inflation during the period. 6 The labor share has not varied much from 1995 to It is 60.4% in 1995 and 59.5% in

16 5 Measuring the Downward Nominal Wage Rigidity using Micro Data In this section we analyze the individual-level data and report some results. For this purpose, we use the KLIPS (Korean Labor and Income Panel Study) from Korea Labor Institute. The KLIPS is a longitudinal survey of household and individuals and its first wave was launched in The data contains the record of each household member s job and income history. First, we plot the distribution of wage changes and examine whether we can observe downward nominal wage rigidity in Korea. Figure 6 shows the histogram of wage changes from 1998 to 2010, respectively. There is a spike around zero for almost all years, and the distribution is skewed to the right. It seems that there exists downward nominal wage rigidity in Korea. To find a more clear answer for this question, we now try to measure the downward nominal wage rigidity following Dickens et al. (2007). From the KLIPS, we calculate the number of workers with wage freezes and cuts for every year, then we report n in Table 1. To identify wage freezes and wage cuts, we select workers who stay in the same job for two consecutive years 7. We calculate the wage changes using the question in the KLIPS which identifies a monthly wage. The distribution of nominal wage changes and the degree of nominal wage rigidity are calculated from this wage growth rate. Figure 7 shows the degree of downward nominal wage rigidity: the x-axis denotes the nominal wage changes 8 (%) and the y-axis shows the fraction of workers (%) for each wage change category. One of the distinctive features is that the distribution is highly condensed around 0. The fraction of workers with wage freezes was over 15% in 2002 and it was over 20% in This means that the fraction of workers with wage 7 We only consider whether they keep the same main job. 8 We consider employees who experience over 50% wage increase as employees with 50% wage changes 16

17 Table 1: Summary of Nominal Wage Changes: Year Sample Freeze Freeze Ratio(f n ) Wage Cuts Cut Ratio(c n ) n , , , , , , , , , , , , , freezes appears to increase over time significantly. This, in turn, resulted in an overall increase in n as shown in Table 1, since those workers experienced nominal wage cuts in the absence of downward nominal wage rigidity, as the measure indicates. Table 1 shows the fraction of workers with wage freezes and cuts along with the degree of nominal wage rigidity (n). The second column reports the number of people in the data who kept a job for two years.the third column shows the number of workers who experience wage freezes and the fourth column reports the ratio of these workers. Similarly, the fifth column and the sixth column report the number of workers with wage cuts and a fraction of those workers, respectively. Finally, the seventh column reports n using the number of workers with wage freezes and cuts. Figure 8 plots some statistics from Table 1 to see if there are any trends. It seems that the overall degree of wage rigidity measured as n tends to increase over time. The fraction of employees with wage freezes (f n ) becomes larger since 2005, and it is also notable that its ratio increases even more after the financial crisis in 2009, as the fraction of workers with wage cuts increases. It provides a stark contrast to the fact that the fraction of workers with wage cuts was relatively large compared to the 17

18 Figure 7: Distribution of Nominal Wage Changes (%): Fraction Fraction Fraction Fraction Fraction Fraction Fraction Fraction Fraction Fraction Fraction Fraction Source: KLIPS 18

19 Figure 8: Downward Nominal Wage Rigidity in Korea Degree of DNWR Fraction of workers with NW freezes Fraction of workers with NW cuts workers with wage freezes in 1999, suggesting that the wage rigidity may become more severe compared to the past. This is contradictory to the findings in Section 4 which show that the average wages declined flexibly in Then, how can we reconcile the findings in this section with the ones from Section 4? We first need to note that there can be discrepancies in the degree of wage rigidity depending on the data. For example, Fuhrer et al. (2012) document that firm level wage distribution from the Occupational Employment Statistics (OES) is quite different from the individual level, indicating the firms ability to adjust its labor costs with more flexiblity than expected. The wage rigidity measured at the individual level only keeps track of employees who stay in the job for at least two consecutive years. Therefore, this overlooks the possibility that a firm can replace workers with cheaper labor costs. The wage rigidity measured at the individual level may be different if a small fraction of workers within a firm keep their jobs when a large fraction of work- 19

20 Figure 9: Changes in Labor Cost: Source: Ministry of Employment and Labor ers is replaced. This is one of the possible explanations that can reconcile these two different findings. We investigate this possibility using the survey data which is called the report on labor cost of enterprise survey from the Ministry of Employment and Labor from 1994 to Figure 9 indicates that the conjecture were indeed the case. Firms seemed to be able to adjust their labor cost flexibly as the data shows. In response to the Asian financial crisis in 1998, the labor cost growth rate fell by 9.7%. We can also observe some instance that their labor cost growth rate dropped significantly even at negative rates in 2000 and Therefore, we suggest that the difference observed in two different sources of data may come from the ability of firms who can change their labor cost when necessary. However, one should be careful in interpreting this explanation as a evidence that there is inconsistency in the wage rigidity at the individual level, rather we consider 20

21 the possibility that firms can flexibly change their labor costs via the compositional changes in workers in the face of negative shocks. We will investigate this issue more deeply in the later section using as much industry level data as is available. Further research would be needed to examine whether firms can change labor costs even with the presence of nominal wage rigidity. 6 The Downward Nominal Wage Rigidity at Industry- Level As shown in Section 5, the nominal wage rigidity may appear different depending on the data that are used. One of the possible explanations is that firms may be able to flexibly change their labor costs by adjusting the composition of workers. Due to the lack of observations of micro data 9, it is hard to evaluate this claim. As an alternative, we investigate the issue at the industry level using the data from the Survey Report on Labor Conditions by Employment Type from the Ministry of Labor and Employment 10. With this data, we calculate the average wage growth rate at the industry level and consider how the number of temporary workers changes over time. Our analysis at the industry level in this section is based on the findings in the existing literature. Kuroda and Yamamoto (2013) note that nominal wages and prices move in a similar direction and determine that this is the cause of deflation in Japan. They find, however, that this phenomenon may not hold up the same across sectors. They also document that nominal wages for the manufacturing sector increased even with the decline in the average price of durable goods. This is in stark contrast to the service sector which experienced wage increases with the increase in service prices. 9 It does not look proper to classify workers into temporary workers and regular workers to answer the question. For example, the fraction of temporary workers was 22.4% in 2002 which is about It is an annual survey at a sample of about 32,000 establishments. 21

22 Those findings suggest that there can be differences in nominal wage rigidities at the macro- and the industry levels. The industry level data mainly comes from the report on the labor force survey from the Ministry of Employment and Labor. It provides basic data collects information on wages and working hours so that we can analyze the wage changes for each establishment and industry. It covers 28,000 sampled establishments, so it has a large number of observations compared to the micro data we use in this paper. Section 3 shows that the overall wage changes seem to be flexible and are made with some lags. Following Kuroda and Yamamoto (2013), we investigate whether the downward nominal wage rigidity can be found at the industry level in Korea also. Figure 10 indicates that the manufacturing sector exhibits higher volatility in its growth rate than the service sector. Indeed, the wage volatility in the manufacturing sector is smaller compared to the service sector. In 1998, the growth rate of manufacturing and service sectors declined to -7.3% and -3.2% and the wage growth rates also grew at the negative rates at -4.3% and -2.2%. However, we see a different picture in The manufacturing sector did not cut wages; rather wages increased even under unfavorable growth rates. Different from the manufacturing sector, the service sector cut its wages by -1.8%. This confirms that the downward nominal wage rigidity may be higher in the manufacturing sector as discussed in Kuroda and Yamamoto (2013). However, wage rigidity at industry levels may be different from those at individual levels for the following reasons. First, when some workers receive high wage increases to offset wage cuts for the rest of the workers, then overall wages may not go down. In this case, averaging the wage growth rate may be misleading. Second, we need to consider the substitutability of the labor force in each industry. For example, let s suppose there is an industry that can change labor composition in response to shocks. With a sudden negative shock, this industry would fire workers with high salaries and 22

23 Figure 10: Wage Changes and Real Growth Rate in Manufacturing and Service Sectors ( ) 25 Nominal wage growth, manufacturing(%) Real GDP growth, manufacturing(%) Nominal wage growth, services(%) Real GDP growth, services(%) Manufacturing Services replace them with cheaper labor. In this case, the overall wage growth rate may go down, but this phenomenon may not be observed when we plot the wage growth using individual level data of workers who remain in the job. It is likely that those workers are not substitutable so they maintain higher wage levels. It is hard to demand wage cuts for these employees. To determine whether these are factors we need to consider in explaining differences in wage rigidity across industries, we plot wage changes using the individual level data and examine the ratio of part time workers in each industry. Figure 11 depicts the measure of the downward nominal wage rigidity(n) for all industries - manufacturing and service sector. Figure 10 tells us that the manufacturing sector tends to have higher wage rigidity than the service sector. Yet, Figure 11 shows the opposite picture: The service sector has higher downward nominal wage rigidity than manufacturing sector. As mentioned, this suggests that wage increases rates are larger for workers who keep their jobs and experience wage increase even though the fraction of workers with wage increases is small. When there is substitutability in workers, in other words when firms are able to 23

24 Figure 11: Wage Rigidities in Manufacturing and Service (n): All industries Manufacturing Services replace workers with cheaper labor, the average wage growth may appear to be flexible. This differs from individual-level data which only tracks down the changes in the wage growth of workers who keep their jobs. To examine if this can be a possible explanation for the differences in downward nominal wage rigidity across industries, we calculate the ratio of part time workers in each industry 11. Because they are temporary workers, one needs to be careful when interpreting the result. In the data, the definition of parttime worker does not exactly equal the definition of non-standard workers. The left panel in Figure 12 shows the ratio of part-time workers in the manufacturing and service sectors from 1993 to The right panel shows their wage changes during this time. It shows that the wage changes for temporary workers are quite volatile as expected compared with regular workers. The manufacturing sector has relatively fewer temporary workers. This tells us why the average wage of manufacturing sector tends to be persistent. With a small fraction of part-time workers, it is difficult to 11 For this, we use the raw data from the Ministry of Employment and Labor data since the number of observations is relatively small to classify workers into industries using the KLIPS. This raw data is not a panel data but it allows us to divide workers whether they are full-time or part-time workers. 24

25 Figure 12: The Ratio of Part-time Workers and Wage Changes in Manufacturing and Service Sectors All industries Manufacturing Services Regular workers Temporary workers change the overall wage flexibly. Downward nominal wage rigidity or wage changes may influence price setting by firms; thus it will ultimately change the overall price levels. We make an index (which is denoted 1991=100) for the nominal wage and consumer price index for each industry. As shown Figure 13, wages appear to be higher in the manufacturing sector even though the price level for the manufacturing sector is low. Based on the equation (2), it is possible that an increase in labor productivity in the manufacturing sector may lead to a decrease in unit labor cost. Another possibility is that the labor share (α) is small in the manufacturing sector so the price level does not vary over time even with the increase in wages. Rather, in the service sector with a higher labor share, the increase in wages contributes to higher price levels. 25

26 Figure 13: Trend of Price Level and Wage for Each Industry Index (1991=100) Nominal wage (total) Nominal wage (goods) Nominal wage (services) CPI (total) CPI (goods) CPI (services) Conclusion This paper documents the existence of the downward nominal wage rigidity using micro and macro data at aggregate and industry levels from 1980 to It reveals that there have been flexible adjustments in wage levels on average with a lag. We think that the lagged adjustment is because the wage contract is done on an annual basis so it does not promptly change in response to a shock. Unlike, the evidence from the macro level data, the evidence based on micro-data suggests that nominal wages are downwardly rigid and the degree increases over time. As already noted, discrepancies arise when firms can change their labor costs through changes in the composition of labor force. Even though we examined whether this might be the possibility given the data available, there should be further research on this issue to answer the questions presented in the paper. We also show that there can be intra-industry differences. Wage volatility in the manufacturing sector is smaller compared to the service sector on average with higher 26

27 downward nominal wage rigidity at the individual level. Based on this evidence, we suggest possible explanations to reconcile the discrepancies between the macro and micro level and between industry and individuals within each industry. The findings in this paper indicate that firms were able to adjust their labor costs in response to economic conditions even with the nominal wage rigidity observed at the individual level. This nominal wage rigidity appears to be greater in Korea compared to other countries, and it increases over time. In this paper, we limit our attention the existence of the downward nominal wage rigidity using descriptive measures so that it requires more precise quantitative analysis to further support the possible explanations mentioned in this paper. 27

28 References George A. Akerlof, William R. Dickens, and George L. Perry. The Macroeconomics of Low Inflation. Brookings Papers on Economic Activity, 27(1):1 76, URL Thomas K. Bauer, Holger Bonin, and Uwe Sunde. Real and Nominal Wage Rigidities and the Rate of Inflation: Evidence from West German Micro Data. IZA Discussion Papers 959, Institute for the Study of Labor (IZA), December URL Mary Colleen Daly, Bart Hobijn, and Brian Lucking. Why has wage growth stayed strong? FRBSF Economic Letter, (apr2):10, URL William T. Dickens, Lorenz Goette, Erica L. Groshen, Steinar Holden, Julian Messina, Mark E. Schweitzer, Jarkko Turunen, and Melanie E. Ward. How wages change: Micro evidence from the international wage flexibility project. Journal of Economic Perspectives, 21(2): , doi: /jep URL Michael W.L. Elsby. Evaluating the economic significance of downward nominal wage rigidity. Journal of Monetary Economics, 56(2): , March URL Ernst Fehr and Lorenz Goette. Robustness and real consequences of nominal wage rigidity. Journal of Monetary Economics, 52(4): , May URL JEFFREY C. Fuhrer, GIOVANNI P. OLIVEI, and GEOFFREY M. B. TOOTELL. Inflation dynamics when inflation is near zero. Journal of Money, Credit and Bank- 28

29 ing, 44:83 122, ISSN doi: /j x. URL Takeshi Kimura and Kazuo Ueda. Downward nominal wage rigidity in japan. Journal of the Japanese and International Economies, 15(1):50 67, ISSN doi: URL Sachiko Kuroda and Isamu Yamamoto. The Impact of Downward Nominal Wage Rigidity on the Unemployment Rate: Quantitative Evidence from Japan. Monetary and Economic Studies, 21(4):57 85, December URL Sachiko Kuroda and Isamu Yamamoto. Wage Fluctuations in Japan after the Bursting of the Bubble Economy: Downward Nominal Wage Rigidity, Payroll, and the Unemployment Rate. Monetary and Economic Studies, 23(2):1 29, May URL Sachiko Kuroda and Isamu Yamamoto. Wage Premiums for Firms Work-Life Balance Practice: Evidence from Japanese matched firm-worker data (Japanese). Discussion Papers (Japanese) 13004, Research Institute of Economy, Trade and Industry (RI- ETI), February URL Sachiko Kuroda and Isamu Yamamoto. Is downward wage flexibility the primary factor of japan s prolonged deflation? Asian Economic Policy Review, 9(1): , ISSN doi: /aepr URL David E Lebow, Saks Raven E, and Wilson Beth Anne. Downward Nominal Wage Rigidity: Evidence from the Employment Cost Index. 29

30 The B.E. Journal of Macroeconomics, 3(1):1 30, October URL Kenneth J. McLaughlin. Rigid wages? Journal of Monetary Economics, 34(3): , December URL Kenneth J. McLaughlin. Are nominal wage changes skewed away from wage cuts? Review, (May): , URL Kenneth J. McLaughlin. Asymmetric wage changes and downward nominal wage rigidity, Stephanie Schmitt-Grohé and Martin Uribe. Downward nominal wage rigidity and the case for temporary inflation in the eurozone. Journal of Economic Perspectives, 27(3): , doi: /jep URL Jennifer C Smith. Nominal Wage Rigidity in the United Kingdom. Economic Journal, 110(462):C176 95, March URL Kengo Yasui and Shinji Takenaka. Deflation and Downward Nominal Wage Rigidity: Evidence from Japan. Discussion Papers in Economics and Business 05-21, Osaka University, Graduate School of Economics and Osaka School of International Public Policy (OSIPP), July URL 30