U NEMPLOYMENT AND U NDEREMPLOYMENT IN THE
P HILIPPINES: THE C ONEXION BETWEEN L ABOR
M ARKETS AND P OVERTY
Background paper for the Philippines Poverty Assessment 2004 .
Jose G. Montalvo 6/15/2004
Introduction
The objective of this document is to analyze what we consider the critical elements of the Philippines’ labor market that have a very important influence on poverty. From this perspective we consider the effect of regional and macroeconomic shocks in the evolution of unemployment and participation rates and the determinants of underemployment. Section 1 considers the case of Philippines in the context of the labor markets of the countries in East Asia and the Pacific. Section 2 analyzes, from a regional perspective, the dynamic effect of labor market shocks on employment, unemployment and participation rates. We relate the differences in the evolution by regions with important institutions of the labor markets and, in particular, minimum wages. The third section analyzes the important phenomenon of the underemployed.
Finally section 4 present the basic conclusions.
1. The Philippines’ labor maker from the Asian perspective.
The labor market of the Philippines has many idiosyncratic characteristics that make it different from the rest of the countries of Eastern Asia and the Pacific. Figure 1 shows that the rate of unemployment of the Philippines economy is systematically higher than the rest of the countries of the region. Over the sample period the unemployment in the Philippines has been close to double the average of the countries of the region. It is interesting to notice the opposite patterns of Malaysia and Indonesia.
While the first country has shown a very significant decrease in the rate of unemployment in Indonesia the opposite has happened, specially after the crisis of 1998.
Another disturbing feature of Philippines’ unemployment is that the proportion of college graduates reaches the 16% (13% among men and 21% among women) and it is increasing.
Figure 1. Unemployment rates.
0.0 2.0 4.0 6.0 8.0 10.0 12.0
1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002
Philippines Malaysia Indonesia Singapore Thailand
Source: Asian Development Bank.
In the explanation of the evolution of unemployment there are elements of supply and demand. The changes in the labor supply are basically related with two components: the evolution of working age population and the participation rates. Figure 2 shows that the growth rate of labor force is very volatile in all the countries. If we look at the average we see that, over the period considered, the average growth rate of the working age population has been 2.7% in Philippines, very similar to the 2.8% of Indonesia and Singapore or the 3.1% of Malaysia. However, in Philippines the growth rate of the working age population before the 80’s was 3.6%. Therefore there has been a clear reduction in the growth rate.
Figure 2. Growth rate of the working age population.
-4.0 -2.0 0.0 2.0 4.0 6.0 8.0 10.0 12.0
1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002
Philippines Malaysia Indonesia Singapore Thailand
The participation rate of the Philippines (figure 3) has shown an increasing trend although with high volatility, mostly related with external shocks.
Figure 3. Participation rate. Philippines.
62.0 63.0 64.0 65.0 66.0 67.0 68.0
1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002
Figure 4 shows the comparison of the participation rate for several Asian countries in 2001. The male rate in Philippines is one of the highest of the countries considered in figure 4. Only Myanmar, Lao, Indonesia and China have a higher male participation rate than Philippines. However the participation rate of women in Philippines is more disappointing.
Figure 4. Participation rates by sex and country.
0.0 10.0 20.0 30.0 40.0 50.0 60.0 70.0 80.0 90.0 100.0
f emale male
The increasing trend of the labor force participation in Philippines during the first part of the 80’s is mostly due to the increasing participation of women which has jumped from 40.13 in 1976 to 47.9% in 1985. The participation rate of women stopped growing by the middle of the 80’s until the beginning of the 90’s. After that the growth was slow and volatile but the trend was clearly upwards until reaching 52,2% in 2001.
By contrast the participation rates of males has been stable around 80%.
Another interesting exercise is to look at the participation rates by age. Figure 5 shows this rates by age of the men. It is noticeable the decrease in the participation rate of the youngest age group induce by the increase in schooling. The same effect can be observed in the sample of women. The participation rate of men between 25 and 54 years old is very high, reaching over the 95% level. For the age group between 55 and 64 years old there is an interesting reduction in the participation rate.
This is even more interesting if we consider the labor participation of women in the same age group. Figure 6 shows that, opposite to what happen with men, the participation rate of women increases in the age group between 55 and 64 years old.
Another interesting fact with respect to the participation rates of women is the apparent absence of cohort effects. Since the increase in participation of women can be traded back to the middle of the sixties it would be reasonable to expect a similar participation rate among women 20 years old or older. This would imply that the cohort effect (being born in a period of relatively high women participation rate) would somehow
compensate the increasing participation rate tied to older ages. However this is not observed in the LFS data.
Figure 5. Participation rates by age. Men.
0 10 20 30 40 50 60 70 80 90 100
15-19 20-24 25-34 35-44 45-54 55-64
1990 1995 2000
Figure 6. Participation rates by age. Women.
0 10 20 30 40 50 60 70
15-19 20-24 25-34 35-44 45-54 55-64
1990 1995 2000
Therefore we can summarize basic statistics of the labor market in Philippines by saying that unemployment rates are higher than in other countries of the region but the participation rate is also higher. The behavior of male and female participation rates is very different for some age groups although it is similar for the youngest group.
Finally there would be interesting to analyze the cohort effects in the participation of women since it has some characteristic features.
Young labor market.
Although in the previous section we have visited the issue of participation rates by age groups and, therefore, we have considered the situation of young worker, we did not make any reference to their unemployment rate. Figure 7a shows the recent evolution of the unemployment rate among young workers in the age group between 15 and 30 years old. The first interesting fact is that the unemployment rate among young workers is double the rate of the general labor force. This is important since this are the workers with the highest level of education which seems to imply that many of their abilities are depreciated because of their high level of unemployment. The problem is particularly important for women, which is also important since they have a higher educational level than men. In fact the proportion of college undergraduates and graduates among the young unemployed has increase from 34.6% in 1998 to 39.5% in 2002.
Figure 7a. Youth unemployment in recent years. Age group from 15 to 30 years old.
14 15 16 17 18 19 20 21 22 23 24
1998 1999 2000 2001 2002
Both sexes Men Women
Source: NSO.
The unemployment situation is even worst if we consider only the age group from 15 to 24 years. In this case, as figure 7b shows, the unemployment rate is close to 3 times the general unemployment rate.
14 16 18 20 22 24 26 28 30
1998 1999 2000 2001 2002
Both sexes Men Women
Source: LFS and NSO.
It is worth noticing that job experience is less and less an antidote against unemployment among young population. Only between 2001 and 2002 the proportion of young people unemployed that have had a previous job jump from 48.3% to 54.7%.
Looking at the methods employed by young unemployed to find a job we observe that the direct approach to the employer is the most common procedure (figure 8) reaching in 2002 the 39.1%. The second most used method of searching for a job was the intervention of friends and family (31.5%). It is worrying the low percentage of young people that search for a job using public employment offices. In 2002 only 4.9%
of young people went to a public office to try to find a job.
Figure 8. Job search method among young unemployed seeking for a job
0 5 10 15 20 25 30 35 40 45
1998 1999 2000 2001 2002
Public agency Private agency Approach employers Friends and relatives Place/ answer ads Others
Source: LFS and NSO.
Finally the number of weeks seeking for a job is, in general, lower than the average value for adults over 30 years old. However the age group between 25 and 30 years old has an average duration of the job search that is higher than the one of adults over 30 years old as shown in figure 9.
Figure 9. Average duration of the job search (weeks).
0 2 4 6 8 10 12
1998 1999 2000 2001 2002
15-19 20-24 25-30 over 30
Source: LFS and NSO.
2. Regional evolutions: labor market institutions and shocks.
2.1. Basic facts
There are important differences in the evolution of labor markets across the regions of Philippines. For instance table 1 shows the participation and unemployment rate of each region. The range of variation in the unemployment rate goes from 4.6% up to 16.2%
which is a huge difference. In fact the variance is 9. The participation rate is less variable across regions but still ranges from 57.6%1 up to 74%.
Table 1. Participation and unemployment rates by region. October 2003
Participation Unemployment
NCR 66.4 16.2
CAR 69.0 7.2
Region I 64.8 10.4
Region II 69.9 4.6
Region III 64.2 11.0
Region IVa 67.0 13.0
Region IVb 69.3 6.2
Region V 67.2 7.1
Region VI 68.2 8.1
Region VII 65.2 11.9
Region VIII 71.8 8.1
Region IX 65.1 5.9
Region X 74.0 6.6
Region XI 68.5 8.9
Region XII 68.9 8.8
ARMM 57.6 5.4
Caraga 69.5 9.1
Source: LFS.
Figure 10. The relationship between regional unemployment and labor market participation.
1 Notice that in this case there is a religious factor related with the labor participation of women.
0.0 2.0 4.0 6.0 8.0 10.0 12.0 14.0 16.0 18.0
55.0 57.0 59.0 61.0 63.0 65.0 67.0 69.0 71.0 73.0 75.0
Figure 10 shows that even considering the Autonomous Region of Mindanao the relationship between unemployment and participation is negative. Eliminating this outlier shows an even steeper negative slope. These data show the interest of the regional dimension in the analysis of labor market interactions and the effect of possibly asymmetric shocks.
There are also important regional differences with respect to productivity. Table 2 shows the productivity of the regions of Philippines in the period 1997-2002 as well as the average. Productivity is measured as the ratio of gross regional domestic product over average employed person (over the four quarterly data).
Table 2. Productivity by regions (at constant prices of 1985)
Region 1997 1998 1999 2000 2001 2002 AVE.
Philippines
32,226 31,814 33,096 35,442 34,357 34,798 33,902
NCR Metro Manila 77,116 75,517 75,722 82,484 80,668 81,867 79,252 CAR Cordillera
35,650 36,714 45,338 47,240 45,639 46,310 44,248 Ilocos Region 16,947 18,347 18,618 19,509 19,445 20,456 19,275
Cagayan Valley 15,024 14,420 18,752 20,277 19,578 19,186 18,443 Central Luzon
33,467 30,420 30,527 32,718 31,875 31,791 31,466
Southern Tagalog
38,074 36,987 35,293 37,154 35,419 36,946 36,360 Bicol Region
13,252 12,956 15,439 16,122 15,678 16,333 15,306 Western Visayas
25,215 25,188 28,209 30,129 29,689 29,326 28,508 Central Visayas
29,462 30,017 31,576 33,992 34,843 34,843 32,970 Eastern Visayas
14,133 14,257 16,019 17,016 15,868 15,586 15,749 Western Mindanao
23,171 23,848 23,740 24,944 23,882 23,632 24,009 Northern Mindanao
33,532 28,258 30,052 32,656 31,254 24,666 29,377 Southern Mindanao
25,229 29,052 28,935 31,213 29,749 42,342 32,261 Central Mindanao
26,464 25,741 24,921 26,926 25,398 20,718 24,741 ARMM
13,043 13,058 12,772 12,791 12,582 12,211 12,683 CARAGA
15,966 14,861 15,562 17,807 16,555 16,376 16,232
Obviously there are many reasons for this large difference in productivity ranging from the industrial mixing of the productive structure of each region to the stock of private, public and human capital.
Labor and productivity regional differences do not stop here. The basic institutions of labor makets (like minimum wages, collective bargaining, etc.) show also important regional differences that can be used to pin down important economic estimates. Table 3 shows the minimum wage (in nominal and real terms) and describes the evolution of purchasing power as a function of the inflation differentials.
.
Table 3. Regional differences in the highest nominal and real minimum wage.
REGION MINIMUM WAGE FOR AS FOR DEC02
FOR JANUARY -
December 2003 FOR THE MONTH -
January 2004 PERCENT CHANGE IN REAL WAGE g/
NOMINAL
WAGE REAL WAGE
(AV. 02) NOMINAL WAGE c/ REAL
WAGE d/ NOMINAL WAGE e/ REAL
WAGE f/
NCR 280.00 165.26 280.00 159.89 280.00 155.90 (2.50)
CAR 190.00 119.17 190.00 116.02 190.00 111.90 (3.55)
I 190.00 116.63 190.00 114.10 190.00 112.69 (1.24)
II 185.00 111.53 185.00 112.35 185.00 111.65 (0.62)
III 228.50 140.67 228.50 136.98 228.50 132.85 (3.02)
IV 237.00 138.54 237.00 136.71 237.00 134.35 (1.73)
V 182.00 102.89 182.00 100.16 182.00 98.11 (2.05)
VI 180.00 113.93 180.00 111.66 180.00 110.02 (1.47)
VII 200.00 112.43 200.00 108.72 200.00 107.35 (1.26)
VIII 188.00 110.41 188.00 108.39 188.00 106.94 (1.34)
IX 175.00 105.13 175.00 103.55 175.00 102.22 (1.28)
X 192.00 113.09 192.00 110.13 192.00 107.80 (2.12)
XI 195.00 121.83 195.00 118.33 195.00 115.45 (2.43)
XII 180.00 113.67 180.00 113.00 180.00 110.84 (1.91)
CARAGA 179.00 109.75 179.00 107.18 179.00 104.62 (2.39)
ARMM 140.00 74.57 140.00 72.31 140.00 71.21 (1.52)
2.2. Regional evolutions: shocks and minimum wages.
The aim of this section is to investigate some regularities and asymmetries in the labor market of Philippines, and to find some implications for the importance of migration across regions within the country or to foreign countries as determinants of the level of employment. The analysis concentrates in the evolution of unemployment and participation rates after a labor demand shock. This is important for poverty since, as it was pointed out in the previous Poverty Assessment, poverty and unemployment are very closely related. In addition a very recent paper by Manlagñit (2004) has noticed that many economically active household members can alleviate poverty.
The reference framework is the econometric specification proposed by Blanchard and Katz (1992)2. The main result from their model is that in response to an adverse shock to labor demand an instantaneous downward adjustment of wages takes place. As a consequence, two driving forces determine the new equilibrium. On one side workers tend to migrate to other regions and/or countries, while firms, attracted by the lower wages, move into the shocked country. Depending on the relative speed and magnitude of the two adjustment processes, the reduction in the employment level can be more or less severe. The analysis is based on quarterly data from the Labor Force Survey, from 1992 to the last quarter of 20023. The unemployment and participation rates by region are calculated using the adjusted weights provided by the NSO in the
2 Blanchard O.J. and Katz L.F., Regional Evolutions, Brooking Papers on Economic Activity, Vol. 1992, No. 1, 1-75
3 Unfortunately we could not use the data from 1988 to 1991 since there were many doubts about their integrity and representativeness. We tried to gather information on the location of each variable in each registry, since the original dictionaries were wrong, but we could not find a sufficiently credible correspondence. Additionally some quarter were missing (
LFS files. The same adjustment is used to aggregate employment across areas in each region.
As a preliminary step we look at some facts which have characterized the Philippines labor market in the last decade and in particular the evolution of regional employment and unemployment in comparison with national figures. A natural question is to check if there is any persistence of employment growth rates. During the last decade Philippines regions did not appear to have a sustained difference in employment growth. The line illustrated in Figure 11 has a non significant slope coefficient, and a very low R2 of 0.16. However we can see that there is a high degree of heterogeneity among regions with some clear improvements (Region 8) and drops (National Capital) Region).
Philippines’ regions experienced in the last decade quite different development patterns along which employment has been very different from region to region. The following graphs give an idea of the evolution of employment in the sample period. We Figure 11 :: Persistence of employment growth (yearly data)
have grouped the eleven regions4 in three main groups depending on the relative performance.
In Figures 12 to 16 we show the evolution of cumulative employment growth across time, where cumulative employment growth is defined as the cumulated change in log employment minus its value at the beginning of the period, and employment for each region is the NSO5-standard employment, calculated in deviation from national mean6.
Even though we don’t see in the data any experience of decisive and strong growth –these are the years of the Asian slowdown-, we can clearly identify four regions displaying an appreciable surge in employed labor force, especially during the last years. These regions (see Figure 12) are Central Luzon (Region 3), CALABARZON and MIMAROPA (Region 4), Mindanao in the south west, and NCR, Manila. Note that the first two are close to each other and are all around the capital Manila while the last is located in the south.
Figure 13 shows the cumulative growth of those regions which performed slightly below their score at the beginning of the sample period. Note that either Ilocos Region (Region 1) in the north west, and the other two regions, Western Visayas and Central Visayas (Regions 6 and 7) display a slightly better performance than their close neighbours, the Cagayan Valley (Region 2) and Eastern Visayas (Region 8) respectively, which we award to the label of Heavy Losers in the next graph. This confirms a general trend toward a fall in employment in Philippines during the nineties.
4 Given that after June 1996 and in 2002 the division of some regions changed, we have aggregated Western Mindanao (Region 9), Northern Mindanao (Region 10), Southern Mindanao (Region 11), Central Mindanao (Region 12), ARMM and CARAGA into one single region, called Mindanao, in order to have a common area before and after the change.
5 We choose the National Statistical Office (NSO) definitions because we have more complete series than those collected according with ILO definitions, which are missing in several quarters of the Labor Force Survey.
6 See note 6.
Figure 12:
Much clearer is the deep fall (see Figure 14) of four regions which appear to have been loosing jobs all along the decade with a particularly severe slow down in October 2000. These are Cagayan Valley (Region 2), in the far north east, Bicol Region (Region 5) and Eastern Visayas (Region 8), in the central Philippines just below the regions around Manila, and CAR.
Figure 13 : :
Finally, to give an idea of the movement in employment by geographic areas, we plot again the series above but now grouped according to the administrative repartition of the country in three main or wide regions: Luzon (Figure 5), in the north, which includes Manila, Visayas (Figure 6), in the centre, and Mindanao (Figure 7), the south.
It appears very clearly that the last decade witnessed the surge of the south, the dramatic fall of the centre and the uncertain performance of the north.
Figure 15 : Figure 14 :
After having had a look at employment, let’s now turn the focus on unemployment. The following graphs are intended to provide some intuition about movements and trend in regional unemployment. Note, however, that we have only 11 Figure 16 :
Figure 17 :
observations, so that any conclusion must be taken with due diligence.
The straight line in Figure 18, with a significant coefficient β=0,758 and an R2=0.76, reveals a noticeable persistence in unemployment rate across regions, that is regions tend to have a quite constant unemployment rate. However, the slope less than one suggests that there is a general trend toward improvement in unemployment. In other words, Philippines’ regions seem to have improved altogether remaining in 2002 at almost the same relative positions they were in 1993.
In Blanchard and Katz’s model the correlation between mean unemployment rates and employment growth rates depends on the relative importance of the underlying sources of growth. It implies that if growth comes from labor demand, a negative correlation should occur between average unemployment and employment growth; the opposite should hold if growth comes from labor supply caused by workers’ migration.
As the authors recognize, this simplifying view is far too simplifying, so that the empirical results are not significant. Indeed, the slope coefficient of the straight line drawn in Figure 9 is not significant and the R2 is 0.05.
.
The persistence of unemployment point to the effect of adverse institutions, or labor market rigidities (Blanchard and Wolfers 2000). We will come back later to this issue.
Figure 19 :
Before attempting to disentangle the effects of shocks to employment on itself, unemployment and participation rates, we should make some considerations about the apparently different regional figures of employment. Indeed, it is important to understand how much of the typical movement in regional employment is common to all regions and how much is, on the contrary, region-specific. To answer this question, we run the following regression for each region:
it t i i
it N
N
where Nit is the logarithm employment rate in region i at time t and Nt is the same figure at the national level. We regress two specifications, using both the employment rate calculated as share of the working age population and as share of the labor force. Tables 4 and 5 report the standard error and the t-test for the betas being equal to 1.
The first result is that in both cases the parameters estimated are highly significant and range between 0.67 and 1.85 in the first table, and between 0.39 and 2.24 in the second table, indicating a clearly positive relation between regional and national employment movement.
To what extent the overall movement explains the regional one, depends on the R2 of the regressions? As we can see in Table 4, much of the year-to-year movement of the regional employment -when measured as a fraction of working age population- seems to be explained by movements in national employment. Nevertheless, when we calculate employment as a fraction of the labor force, we get a quite different result:
although the direction of the relation is the same, it appears to be much more volatile across regions and to be much weaker than before –the R2 is considerably lower in all the regressions. This finding may well be due to migration from regions to abroad during bad times instead of internal migration (between regions): indeed, a negative shock to a region’s economy determining a loss of labor force moving from the region to abroad –and an analogous loss of employed people- has a much higher impact over a region’s economy than over the national overall figures. On the other hand, when dealing with working age population, this effect is much lower because the working age population is a bigger figure and only a part of it is willing to migrate. So its movements are proportionally smaller and the relation between regional and national movements appears to be less strict.
A third point to be noted is that for five regions in the first case and four in the second –Region 7 being the only one common to both regressions- the null of a one-to- one comovement cannot be rejected so that we shall consider them moving in line with the overall performance of Philippines. This is actually another piece of evidence in favour of the view sketched above of a country performing badly in a quite uniform recessive phenomenon during the last decade.
2.3. Employment, Unemployment and Participation
The model developed Blanchard and Katz (1992) points out that there can be two adjustment mechanisms in response to an adverse shock in labor demand. One is immigration of firms and the other is the out-migration of labor. The long-run effect on employment depends on the relative strength and speed of these two effects. Therefore
Table 1. Regressions Relating Regional Employment Growth to National Employment Growth (employment as share of working age population)
reg1 reg2 reg3 reg4 reg5 reg6 reg7 reg8 NCR CAR Mindanao
Coeff (β) 0.826** 1.556** 0.696** 0.509** 1.187** 1.854** 1.058** 1.395** 0.857** 0.875** 1.000**
(0.091) (0.130) (0.107) (0.053) (0.069) (0.088) (0.086) (0.140) (0.086) (0.105) (0.031) Constant 0.000 0.001 0.000 -0.001 0.000 0.001 -0.001 0.000 0.000 -0.000 0.000
(0.002) (0.003) (0.002) (0.001) (0.001) (0.002) (0.002) (0.003) (0.002) (0.002) (0.001)
R-squared 0.70 0.80 0.55 0.72 0.89 0.93 0.81 0.74 0.74 0.66 0.97
Test F (β=1) 3.68 18.26 8.02 85.59 7.35 95.06 0.46 7.92 2.78 1.41 0.00
Prob > F (0.06) (0.00) (0.01) (0.00) (0.01) (0.00) (0.50) (0.01) (0.10) (0.24) (1.00) Standard errors in parentheses
*significant at 5%; **significant at 1%
Observations: 37 for each regression
Table 2. Regressions Relating Regional Employment Growth to National Employment Growth (employment as share of labour force)
reg1 reg2 reg3 reg4 reg5 reg6 reg7 reg8 NCR CAR Mindanao
Coeff (β) 0.571* 1.676** 0.716** 0.908** 1.532** 0.582* 0.862** 1.545** 0.391* 2.237** 1.385**
(0.214) (0.222) (0.162) (0.122) (0.174) (0.236) (0.248) (0.270) (0.186) (0.338) (0.110) Constant -0.001 -0.001 0.001 -0.000 -0.001 0.001 -0.001 0.001 0.001 -0.002 0.000
(0.002) (0.002) (0.002) (0.001) (0.002) (0.002) (0.002) (0.003) (0.002) (0.003) (0.001)
R-squared 0.17 0.62 0.36 0.61 0.69 0.15 0.26 0.48 0.11 0.56 0.82
Test F (β=1) 4.00 9.30 3.07 0.57 9.33 3.15 0.31 4.08 10.76 13.37 12.26
Prob > F (0.05) (0.00) (0.09) (0.45) (0.00) (0.08) (0.58) (0.05) (0.00) (0.00) (0.00) Standard errors in parentheses
*significant at 5%; **significant at 1%
Observations: 37 for each regression
our goal is to estimate the long-run effect of a negative shock on labor demand on level of employment, unemployment rate and participation rate and thus measuring the implied effect on migration.
Following these ideas we estimate a log linear system in the following three variables: ei7is the first difference of the logarithm of employment in region i, minus the first difference of the logarithm of Philippines aggregate unemployment. The second,lei, is the logarithm of the ratio between employment to the labor force in the region i, minus the same variable for the entire Philippines. The third lpi is the logarithm of the ratio between the labor force to the working age in region i in deviation from the national average. From the behaviour of these three variables we can easily compute the effect on the level of employment, the unemployment rate and the participation rate8. Our estimated model is:
ipt t
i i
t i i
t i i
i it
iut t
i i
t i i
t i i
i it
iet t
i i
t i i
t i i
i it
lp L le
L e
L lp
lp L le
L e
L le
lp L le
L e
L e
1 , 33 1 , 32 ,
31 30
1 , 23 1 , 22 ,
21 20
1 , 13 1 , 12 1 , 11
10
) ( )
( )
(
) ( )
( )
(
) ( )
( )
(
We allow for two lags of each variable.9 Our specification, imposing that current changes in ei affect both lei and lpi, but not vice versa, and the interpretation of
ie as an innovation to labor demand, reflects the assumption that unexpected movements in employment within the quarter reflect movements in labor demand.
As we can see in the impulse response graphs, in general a negative shock to labor demand produces the following effects: immediately after the shock, the participation rate decreases, the unemployment rate increases and the level of employment drops. In Table 6, we have computed the implied effect on net out migration of a reduction of employment of 1 worker. In the first year, on average10, a
7 Given that we don’t have the level of unemployment we obtained this variable in the following way:
1 1 , 1
1 1
, 1
, ln ln
ln ln
ln ln
t t i t
t t
i t i t
it t
t it
it
i W
W W
E W
E W
W W
E W
e E , where W and it E are it
the working-age population and the number of employee in state i, in time t, respectively and the variable without the subscript i represent the corresponding national aggregation.
8 Indeed we can compute d(U/L)=(E/L)(dln(L/E)) and d(L/W)=(L/P)(dln(L/P)), where U, L, E, and W are unemployment, the labor force, employment, and working age population, respectively. The mean value for the sample of E/L is 0.9019 and for L/P is 0.6906.
9 We have performed the test for choosing the best lag specification, giving different results, between one and free lags, for all the regions.
10 We used the share of population in each single region over the total population as weights.
reduction of 1 worker leads to an increase in unemployment of 0.08 units, a decrease in participation of 0.47 workers and an implied migration of 0.44 workers.
The effect on participation and unemployment rate declines and finally disappears after five to eight quarters. On the other hand the effect on the level of employment reaches on average a plateau of -0.49.
However we note sensible differences across regions: in the first period we note that the impact on migration is much lower in the National Capital Region than in the other regions, while there is a wide variation in the “long-run” effect on the level of employment: indeed we pass from a value of -1.73 in region 2 to values not significantly different from zero in other regions.
Table 6: Implied effect on out-migration in the first quarter of a reduction of 1 worker
Region Out of labor force
Unemploymen t
Implied Out Migration
Region 1 0.67 0.18 0.15
Region 2 0.11 -0.04 0.93
Region 3 0.59 0.22 0.18
Region 4 0.04 -0.26 1.22
Region 5 -0.39 -0.15 1.54
Region 6 0.26 0.41 0.33
Region 7 0.64 0.23 0.12
Region 8 0.91 0.00 0.09
Mindanao 0.73 -0.02 0.29
NCR 0.65 0.34 0.01
CAR 0.74 0.11 0.15
Average5 0.47 0.08 0.44
Figure 12 presents the dynamic impact of a negative shock in the demand of labor on employment growth, participation rates and unemployment. With one exception all the regions behave as expected by the model. The negative shock increases unemployment and decreases participation contemporaneously. Employment is also negatively affected. Over time the unemployment and the participation rates come back to their previous levels while employment remain below its initial level.
However the most interesting effect is the differential dynamic behaviour of the regions. It is quite interesting to notice that in regions where the minimum wages are
Source: Authors calculations based on the impulse response functions.
* See note 8
high the recovery of the unemployment rate takes very long (more than 8 quarters) while in regions where the minimum wages are low this effect is seen in 2 or 3 quarters.
These finding are consistent with the findings in Blanchard and Wolfers (2000). They argue that the evolution of unemployment in Europe cannot be explained only by the effect of shocks or labor market rigidities. Blanchard and Wolfers (2000) focus in the interaction between adverse shocks and adverse market institutions. Some institutions have an important effect on the impact of shocks on unemployment. If minimum wages are too high then an increase in unemployment due to an adverse shock reduces the pressure of unemployment on wages. This would slow the return to lower unemployment. This is exactly what we observe in graph 12.
One would expect to see different dynamic evolution to negative shock in function of the minimum wages of each region. As we saw in table 3 there is a large variability in minimum wages across regions in Philippines. The most dramatic behaviour observed in unemployment and employment after a negative shocks happens in NCR where, at the same time, we observe the highest minimum wage (280 pesos).
The other region that shows a very slow recovery of unemployment and participation rates is Central Luzon which present also one of the highest levels of minimum wages (228.5 pesos). Let’s consider the case of NCR. Is 280 pesos/day a high minimum wage?
One possible way to answer this question is to compare it with other countries. For instance in the US the federal minimum wage is 5.15 dollars per hour. Therefore the minimum daily wage is 0.09% the median family income per year (assuming 8 hours of work per day). In NCR the daily minimum wage is 4.96 dollars. That means 0,2% of the median family income per year, which is more than double the value of the United States.
How does this minimum wage compare with other minimum wages across South Asia and the Pacific area? The minimum wage in NCR is clearly above the minimum wage in China, Vietnam, Indonesia and Thailand. It is also a little higher than the minimum wage in Malaysia and it is only below the value of South Korea and Singapore.
From the region NCR in graph 12 we see that the unemployment rate initially increases because of the negative shock and it takes between 6 and 7 quarters to recover.
Employment takes also a long time to recover but, opposite to what happen in other regions, after two years it has almost recover the pre-shock level. In Luzon we observe something similar, although it takes even longer to recover the pre-shock unemployment
rate. In addition the employment do not recover the pre-shock level.
Variable
step Response (%) S.E.(%) Response (%) S.E.(%) Response (%) S.E.(%) Region1
0 0.00 0.0000 0.00 0.0000 0.00 0.0000
1 -1.69 0.2048 -0.76 0.1381 0.40 0.1432
2 -1.16 0.2995 -0.44 0.1610 0.19 0.1404
3 -0.62 0.3179 -0.21 0.1585 0.08 0.1351
4 -0.41 0.3224 -0.14 0.1514 -0.12 0.1255
5 -0.37 0.3107 -0.09 0.1343 -0.10 0.1026
6 -0.30 0.2507 -0.02 0.0934 -0.08 0.0730
7 -0.27 0.2034 0.01 0.0663 -0.05 0.0520
8 -0.28 0.1720 0.02 0.0474 -0.03 0.0422
9 -0.29 0.1485 0.02 0.0323 -0.01 0.0313
10 -0.30 0.1351 0.01 0.0237 0.00 0.0240
11 -0.31 0.1296 0.01 0.0180 0.00 0.0178
12 -0.32 0.1292 0.00 0.0134 0.00 0.0126
Region2
0 0.00 0.0000 0.00 0.0000 0.00 0.0000
1 -2.46 0.2988 -0.62 0.1690 0.15 0.1953
2 -1.56 0.4628 -0.32 0.2152 -0.34 0.2029
3 -1.92 0.5362 -0.18 0.2371 0.14 0.1922
4 -1.83 0.4751 -0.20 0.1953 -0.04 0.1265
5 -1.81 0.4971 -0.13 0.1624 0.02 0.0729
6 -1.80 0.4770 -0.12 0.1344 0.01 0.0442
7 -1.78 0.4775 -0.09 0.1148 0.01 0.0265
8 -1.77 0.4767 -0.08 0.0987 0.01 0.0183
9 -1.75 0.4757 -0.06 0.0860 0.00 0.0160
10 -1.74 0.4766 -0.05 0.0753 0.00 0.0127
11 -1.74 0.4768 -0.04 0.0661 0.00 0.0110
12 -1.73 0.4774 -0.03 0.0581 0.00 0.0092
Region3
0 0.00 0.0000 0.00 0.0000 0.00 0.0000
1 -1.48 0.1796 -0.60 0.1339 0.40 0.1629
2 -0.89 0.2593 -0.25 0.1475 0.39 0.1628
3 -0.71 0.2100 -0.28 0.1125 0.24 0.1236
4 -0.75 0.2176 -0.27 0.1094 0.26 0.1046
5 -0.66 0.2448 -0.23 0.1198 0.22 0.1081
6 -0.60 0.2661 -0.21 0.1258 0.19 0.1104
7 -0.55 0.2810 -0.19 0.1296 0.17 0.1102
8 -0.50 0.2920 -0.17 0.1314 0.16 0.1089
9 -0.46 0.2997 -0.15 0.1317 0.14 0.1072
10 -0.43 0.3040 -0.14 0.1305 0.12 0.1050
11 -0.39 0.3056 -0.12 0.1282 0.11 0.1023
12 -0.36 0.3051 -0.11 0.1250 0.10 0.0994
Region4
0 0.00 0.0000 0.00 0.0000 0.00 0.0000
1 -1.46 0.1771 -0.39 0.1448 -0.18 0.1397
2 -1.46 0.2776 -0.34 0.1412 -0.21 0.1266
3 -1.06 0.3414 0.15 0.1452 0.15 0.1286
4 -0.91 0.3201 0.03 0.0920 0.00 0.0801
5 -1.01 0.2858 -0.02 0.0588 0.01 0.0482
6 -1.04 0.2955 -0.05 0.0388 -0.03 0.0328
7 -1.01 0.3164 0.00 0.0270 0.02 0.0212
8 -0.98 0.3268 0.00 0.0216 0.01 0.0165
9 -0.98 0.3276 0.00 0.0136 0.01 0.0122
10 -0.98 0.3291 0.00 0.0098 0.00 0.0091
11 -0.98 0.3326 0.00 0.0057 0.00 0.0058
12 -0.97 0.3351 0.00 0.0040 0.00 0.0040
Impulse responses to an Employment exogenous shock
Employment Participation Rate Unemployment Rate
Variable
step Response (%) S.E.(%) Response (%) S.E.(%) Response (%) S.E.(%) Region5
0 0.00 0.0000 0.00 0.0000 0.00 0.0000
1 -2.07 0.2509 -0.12 0.1545 -0.08 0.1294
2 -1.53 0.3913 0.06 0.145 -0.18 0.1431
3 -1.84 0.4731 -0.01 0.1443 -0.06 0.1522
4 -1.58 0.4401 0.06 0.0802 -0.08 0.106
5 -1.68 0.4471 0.02 0.0752 -0.05 0.0885
6 -1.57 0.4267 0.04 0.0575 -0.05 0.0698
7 -1.60 0.4375 0.02 0.0491 -0.04 0.0599
8 -1.54 0.4340 0.03 0.0418 -0.03 0.0507
9 -1.54 0.4445 0.02 0.0350 -0.03 0.0443
10 -1.52 0.4474 0.02 0.0308 -0.02 0.0388
11 -1.51 0.4558 0.01 0.0261 -0.02 0.0342
12 -1.50 0.4604 0.01 0.0233 -0.02 0.0304
Region6
0 0.00 0.0000 0.00 0.0000 0.00 0.0000
1 -1.80 0.2177 -0.41 0.1496 0.76 0.2078
2 -1.01 0.3610 -0.23 0.1606 0.27 0.2367
3 -0.66 0.4355 -0.18 0.1713 -0.01 0.2318
4 -0.49 0.4568 -0.12 0.1385 -0.06 0.1757
5 -0.38 0.4413 -0.02 0.0974 0.00 0.0859
6 -0.29 0.4260 0.01 0.0632 -0.01 0.0533
7 -0.28 0.4158 0.00 0.0404 -0.02 0.0301
8 -0.30 0.4065 0.00 0.0229 0.00 0.0228
9 -0.30 0.3992 0.01 0.0111 0.00 0.0138
10 -0.30 0.3944 0.00 0.0072 0.00 0.0094
11 -0.30 0.3915 0.00 0.0040 0.00 0.006
12 -0.31 0.3899 0.00 0.0030 0.00 0.0031
Region7
0 0.00 0.0000 0.00 0.0000 0.00 0.0000
1 -1.94 0.2358 -0.83 0.129 0.55 0.1273
2 -0.83 0.3546 -0.24 0.1655 0.32 0.1475
3 -0.45 0.3548 -0.34 0.1716 0.05 0.146
4 -0.30 0.3168 -0.23 0.1559 0.02 0.1224
5 -0.35 0.2789 -0.28 0.1284 0.02 0.0974
6 -0.15 0.2565 -0.14 0.1282 0.05 0.0721
7 -0.09 0.2392 -0.15 0.1212 0.03 0.0595
8 -0.02 0.2243 -0.10 0.1166 0.02 0.0498
9 0.01 0.2023 -0.09 0.1034 0.02 0.0423
10 0.06 0.1842 -0.07 0.0943 0.01 0.0355
11 0.09 0.1714 -0.06 0.0837 0.01 0.03
12 0.12 0.1620 -0.04 0.0743 0.01 0.0252
Region8
0 0.00 0.0000 0.00 0.0000 0.00 0.0000
1 -1.69 0.2044 -0.99 0.2184 0.15 0.2546
2 -0.59 0.3231 0.00 0.2197 0.42 0.2235
3 -0.83 0.3095 -0.35 0.2177 0.08 0.2082
4 -1.08 0.3199 -0.22 0.1896 0.17 0.1693
5 -1.06 0.3557 -0.14 0.1943 0.17 0.1470
6 -1.07 0.3801 -0.10 0.1758 0.12 0.1327
7 -1.17 0.4193 -0.09 0.1725 0.10 0.1160
8 -1.21 0.4636 -0.06 0.1605 0.09 0.1020
9 -1.23 0.4996 -0.04 0.1494 0.07 0.0898
10 -1.27 0.531 -0.03 0.1337 0.06 0.0760
11 -1.30 0.5617 -0.02 0.1195 0.05 0.0652
12 -1.31 0.5872 -0.01 0.1048 0.04 0.0561
(Cont'd) Impulse responses to an Employment exogenous shock
Employment Participation Rate Unemployment Rate