Unemployment | Krisna 'imed' Gupta

Indonesian consumption growth, labor share, and saving rate

Sun, 23 Mar 2025 12:06:59 +0700

Recently, there have been many reports about the weakening of Indonesia’s middle class. The middle class in Indonesia has been shrinking since 2018 and now makes up only 17% of the total population. However, consumption growth often relies on the middle class. Unsurprisingly, Indonesia’s consumption growth has long lagged behind that of other countries in the region, except Thailand.

Indonesia’s consumption growth has been below the general pace of economic growth for quite some time. How is it possible for economic growth to outpace consumption growth?

One explanation is that this economic growth benefits those with a low consumption share. Consumption share refers to the portion of an individual’s income spent on consumption. Typically, this group consists of wealthy individuals or corporations. These entities already have high consumption levels. Any additional income is more likely to be allocated toward increasing assets (savings) or spending abroad. I previously discussed this issue in The Conversation Indonesia.

I will attempt to explore this theory by presenting two points: (1) Indonesia’s saving rate is increasing, and (2) the labor share is decreasing. I illustrate these two points using data from Total Economy - Integrated Institutional Sector Accounts (trillion rupiahs), 2016 - 2023 from BPS. This balance sheet is a very useful and interesting dataset as it includes calculations from all three methods of calculating Gross Domestic Product (GDP). I have made slight adjustments to the data, which can be viewed at the end of this post.

Increasing Saving Rate

The saving rate reflects the proportion of income saved by the population. A high saving rate indicates that people save more than they spend on consumption. As discussed above, the economic balance sheet does not differentiate the owners of the saving rate, but a significant increase in the saving rate typically signals a rising income share among groups more likely to save, such as wealthy individuals or corporations.

import seaborn as sns
import matplotlib as plt
import pandas as pd

data=pd.read_excel("neraca.xlsx",sheet_name="Sheet3")
dat=data[["tahun","LS","KS"]]
dat=dat.melt("tahun",var_name="jenis",value_name="nilai")
det=data[["tahun","srhh"]]

ax=sns.lineplot(data=det,x="tahun",y="srhh")
ax.set(xlabel="",ylabel="(%)",title="Saving rate (i.e., disposable income minus consumption),\nfrom Neraca Ekonomi BPS")

[Text(0.5, 0, ''),
Text(0, 0.5, '(%)'),
Text(0.5, 1.0, 'Saving rate (i.e., disposable income minus consumption),\nfrom Neraca Ekonomi BPS')]

Since 2016, Indonesia’s saving rate has continued to increase. COVID-19 forced everyone to reduce their savings, even increasing debt to maintain consumption. However, since then, the saving rate has risen again, even higher than usual. It seems that COVID-19 really dealt a blow to Indonesian consumers.

Some analysts say that one of the problems is weak wage growth. Reed, Pasha, and Gonzales, in their report for the World Bank Group, wrote about weak wage growth and profits in manufacturing companies with high foreign ownership. The image excerpt below is from the report.

In the economic balance sheet, we can also calculate how much primary income is allocated for various purposes. I divide it into three: taxes, employee expenses, and profits. Profits consist of income that has already been reduced by taxes and employee expenses. I’ve plotted this below, but I excluded taxes as the amount is very small, around 5% of the total balance sheet¹.

ax=sns.lineplot(data=dat,x="tahun",y="nilai",hue="jenis",palette="tab10")
ax.set(xlabel="",ylabel="(%)",title="Labor share (LS) and profit share (KS) of income, from Neraca Ekonomi BPS")

[Text(0.5, 0, ''),
Text(0, 0.5, '(%)'),
Text(0.5, 1.0, 'Labor share (LS) and profit share (KS) of income, from Neraca Ekonomi BPS')]

And it’s quite apparent that there has been progressive growth in the share of income absorbed by companies. As hypothesized above, an increase in the income share for companies is more likely to be saved, while an increase in the income share for employees is more likely to be consumed. So, at least based on this result, it makes sense.

Anyway, this post is actually just to share the chart. We can have a discussion outside of this post. Let me know what you think.

Below is the data I used. All of it was downloaded from Total Economy - Integrated Institutional Sector Accounts (trillion rupiahs), 2016 - 2023 with slight adjustments.

data

	tahun	L	K	T	va/output	LS	TS	KS	srhh
0	2016	40.576885	55.127270	4.295845	52.877975	41.728750	4.435895	53.835356	30.864658
1	2017	41.472019	53.934836	4.593145	53.108782	42.704020	4.746633	52.549347	31.795110
2	2018	39.977940	55.237447	4.784614	52.986106	41.343331	4.965863	53.690807	32.159447
3	2019	39.252272	55.950651	4.797077	52.971298	40.403369	4.954449	54.642181	31.793236
4	2020	37.304707	58.410155	4.285138	53.170697	38.341668	4.418889	57.239443	30.021169
5	2021	37.283856	58.021987	4.694157	53.484228	38.272325	4.833429	56.894245	33.807914
6	2022	35.453018	59.557783	4.989199	53.799988	36.094115	5.095040	58.810845	38.504459
7	2023	34.970118	60.019108	5.010774	53.400921	36.068580	5.184493	58.746927	36.506632

This 5% figure is quite interesting because Indonesia’s corporate income tax rate is much higher than that. This could make for an entire post on its own—maybe another time. ↩︎

Checking BPS open unemployment rate data

Sat, 18 Jan 2025 14:06:59 +1100

Some time ago, my colleague Anton and I were invited to chat on the Indah G Show platform. It was a casual, mostly spontaneous conversation. Unfortunately, in these impromptu settings, I often don’t have data memorized, and there’s no opportunity to check data – or at least doing so would break the flow. Quite different from chatting on Twitter, which is asynchronous.

One such moment was when we were discussing the unemployment rate. At the time (the recording was around October or November 2024), news about deflation and unemployment was widespread, and many were claiming unemployment was high. I said that the unemployment rate at the time probably wasn’t that different from previous periods. I also mentioned that unemployment rates can vary by region. Unfortunately, checking provincial unemployment figures takes a bit of time.

Now it’s time to check the data. I claimed that unemployment was probably not higher than usual in 2024, and that it could be high in Aceh but low in Jakarta, so it’s not so bad (or at least not worse than typical).

I use the open unemployment rate by province table from BPS. BPS uses the Sakernas survey conducted in two different months: February and August. I plot the last 10 years, separately for February and August. I select a few provinces: Aceh and Jakarta (since I mentioned these on the podcast), plus Yogyakarta, Bali, and North Maluku (no particular reason). The data is shown below.

import seaborn as sns
import matplotlib as plt
import pandas as pd

data=pd.read_excel("ur.xlsx")
feb=data.query("bulan=='Februari'")
aug=data.query("bulan=='Agustus'")

ax=sns.lineplot(data=feb,x="tahun",y="ur",hue='daerah',palette='Paired')
sns.move_legend(ax,loc='lower right',bbox_to_anchor=(1.35,.5))
ax.set(xlabel="",ylabel="Unemployment rate (%)",title="Annual unemployment rate, February")

[Text(0.5, 0, ''),
Text(0, 0.5, 'Unemployment rate (%)'),
Text(0.5, 1.0, 'Annual unemployment rate, February')]

ax=sns.lineplot(data=aug,x="tahun",y="ur",hue='daerah',palette='Paired')
sns.move_legend(ax,loc='lower right',bbox_to_anchor=(1.35,.5))
ax.set(xlabel="",ylabel="Unemployment rate (%)",title="Annual unemployment rate, August")

[Text(0.5, 0, ''),
Text(0, 0.5, 'Unemployment rate (%)'),
Text(0.5, 1.0, 'Annual unemployment rate, August')]

Obviously, 2020 saw the COVID-19 pandemic, which caused unemployment rates to spike. Bali and Jakarta experienced far worse problems than other regions and the national average (red line), which makes sense given that both are highly service-oriented economies.

Overall, we can see that by August and February 2024, unemployment rates had largely returned to the pre-pandemic trend. Moreover, Jakarta has historically had a high unemployment rate – above the national average – just like Aceh. In fact, Jakarta’s unemployment rate is often higher than Aceh’s.

In short:

(1) the unemployment rate is probably not as bad as the news suggests, since it’s approaching the 10-year trend and has generally been improving since the COVID disruption. (2) Jakarta has an unemployment rate comparable to Aceh’s, and both are above the national average.

It’s also important to note that the unemployment rate alone is insufficient for labor market analysis, as many other indicators need to be considered. First, it’s important to know the labor force size, which can rise due to various factors (e.g., people who normally continue to university decide not to, thus entering the labor force), or even fall (people become so desperate about finding work that they give up and exit the labor force). Then there are hours worked (BPS defines someone as employed if they work at least 1 hour per week, so someone could be working but with reduced hours, indicating a weak labor market). And most importantly, the wages paid for those hours. I previously discussed the potential impact of minimum wages on the labor market here.

In other words, it’s quite possible that the unemployment rate hasn’t changed much, but what’s actually happening is a reduction in hours worked or wages. This is like when snack prices stay the same but the contents shrink.

Analyzing the labor market is not straightforward, which is why even among economists, labor economics is a whole specialization. Hopefully this helps you be more critical when you see these numbers going forward.

By the way, BPS has started disseminating more wage data – a very welcome development. Here’s one example of wage data on the BPS website. Also be careful looking at averages, since averages will be skewed and inflated by high earners, to some degree.

Hope this post helps clarify things.

## This is a copilot generated code.
import pandas as pd
from io import StringIO

# Tab-delimited data
data = """daerah\t2015\t2016\t2017\t2018\t2019\t2020\t2021\t2022\t2023
ACEH\t11226\t13627\t14809\t13814\t15065\t18099\t17037\t16772\t17585
DKI JAKARTA\t17012\t23181\t23826\t25238\t25236\t28420\t30662\t32685\t42354
DI YOGYAKARTA\t10440\t12070\t12281\t12554\t13275\t15771\t15098\t14916\t16478
BALI\t11038\t14852\t15624\t15889\t16408\t17775\t17662\t16857\t18521
MALUKU UTARA\t13607\t15226\t15760\t15864\t17425\t23338\t21131\t18278\t27078
INDONESIA\t11434\t14068\t14731\t15275\t15823\t17696\t18089\t17542\t19027"""

# Use StringIO to read the string data into a pandas dataframe
df = pd.read_csv(StringIO(data), delimiter='\t')
df_long = pd.melt(df, id_vars=['daerah'], var_name='year', value_name='value')

## Seaborn
ax=sns.lineplot(data=df_long,x="year",y="value",hue='daerah',palette='Paired')
sns.move_legend(ax,loc='lower right',bbox_to_anchor=(1.35,.5))
ax.set(xlabel="",ylabel="Rp",title="Hourly wage by province, Rupiah/hour")

[Text(0.5, 0, ''),
Text(0, 0.5, 'Rp'),
Text(0.5, 1.0, 'Hourly wage by province, Rupiah/hour')]

Unemployment | Krisna 'imed' Gupta

Indonesian consumption growth, labor share, and saving rate

Increasing Saving Rate

Declining Labor Share

Checking BPS open unemployment rate data