LinkedIn Hiring Rate Analysis: Labor Market Dynamics (2018-2025)#
The analysis of LinkedIn Hiring Rate (LHR) data reveals distinct patterns across Middle East & North Africa, Afghanistan & Pakistan (MENAAP) economies, with notable variations in labor market recovery. The countries analysed are Algeria, Bahrain, Egypt, Jordan, Morocco, Qatar, Saudi Arabia, Tunisia, United Arab Emirates
LinkedIn Hiring Rate Insights#
General Note#
To interpret these results appropriately, it’s important to consider LinkedIn’s guidance on the level of disaggregation at which the hiring rate estimates are valid:
For Algeria, Bahrain, Egypt, Jordan, Morocco, Qatar, Saudi Arabia, and Tunisia, LinkedIn deems the hiring rate valid across tech-related industries.
For the United Arab Emirates, estimates are considered valid across a broader set of industries.
For Iraq and Pakistan, LinkedIn only considers the estimates valid when presented at the national level.
The following insights are presented at the national level and are therefore valid for all countries.
COVID-19 Impact (2020)#
All markets experienced a significant contraction in 2020.
Uniform negative growth rates were observed across countries (~-0.65%).
Morocco (-0.8045%) and Tunisia (-0.7936%) recorded the largest declines.
Recovery patterns began to diverge in magnitude, though they followed similar overall trends.
Post-COVID Recovery Surge (2021)#
A sharp increase in April 2021 hiring rates is visible across countries, driven by year-over-year (YoY) calculations (see methodology).
This spike is primarily due to base effects from April 2020, when hiring plummeted.
The surge reflects a rebound in labor market activity, not necessarily a hiring boom.
Peak YoY growth rates in mid-2021:
Morocco: 3.503% (April)
Algeria: 3.4545% (April)
Tunisia: 3.118% (April)
Pakistan: 2.48% (May)
Iraq: 2.1892% (April)
UAE: 2.0987% (April)
Qatar: 2.0857% (April)
Jordan: 2.0499% (April)
Saudi Arabia: 1.9015% (April)
Egypt: 1.4289% (April)
Bahrain: 1.43% (May)
Most countries peaked in April, except Pakistan and Bahrain, which peaked in May.
The synchronized peaks suggest a regional “catch-up” effect after COVID-19 disruptions.
Market Stabilization Patterns (2022–2024)#
Morocco: Continued growth through 2022 until December, followed by mostly negative or near-zero growth.
Algeria: Stable trend around zero; slightly negative since December 2023.
Tunisia: Positive growth in 2022 with a May peak (0.91%); mostly negative growth from April 2023 onward.
Pakistan: Positive growth throughout 2022 until October 2023, then fluctuating around zero with short periods of positive/negative growth.
Iraq: Mild positive growth during 2022; slight negative or near-zero growth through 2023–2024 with a July 2024 peak (0.12%), then stable around zero.
United Arab Emirates: Positive growth during the first three quarters of 2022, followed by slightly negative values near zero.
Qatar: Dynamic trend with early 2022 growth, a dip in December 2022 (-0.04%), followed by a peak in December 2023 (+0.47%), and mostly negative but near-zero values afterward.
Jordan: Positive growth during 2022; persistent negative values from January 2023 to September 2024 (except May 2023), followed by slight improvement post-October 2024.
Saudi Arabia: Fluctuating around zero since 2022; notable variations include: May 2022: +0.45%, April 2023: -0.27%, June 2023: -0.26%, July 2023: +0.36%.
Egypt: Some growth in early 2022; mostly negative from late 2022 to April 2024. Beginning July 2024, stable near-zero values with slight improvements.
Bahrain: Some growth in 2022, but almost all periods since January 2023 show negative growth.
Exploring relationship between LHR and conflict#
Correlation Between Conflict Fatalities and LinkedIn Hiring Rate#
We explored the relationship between conflict-related fatalities and the LinkedIn Hiring Rate (YOY) across several countries using monthly data.
Overall, we do not observe a strong or consistent negative correlation — that is, higher fatalities do not clearly correspond to lower hiring growth.
Country-Level Pearson Correlation Coefficients
Country |
Pearson r |
Interpretation |
|---|---|---|
Egypt |
+0.24 |
Weak positive correlation |
Iraq |
+0.23 |
Weak positive correlation |
Morocco |
+0.18 |
Very weak positive correlation |
Saudi Arabia |
+0.19 |
Very weak positive correlation |
Tunisia |
+0.08 |
Very weak positive correlation |
Pakistan |
-0.12 |
Very weak negative correlation |
Jordan |
-0.17 |
Very weak negative correlation |
Algeria |
-0.23 |
Weak negative correlation |
None of the countries show a strong negative correlation. In fact, most exhibit weak or even positive correlations, suggesting that conflict severity (as measured by fatalities) may not be a reliable predictor of changes in hiring rate based on LinkedIn data.
Correlation Between Conflict Number of Events and LinkedIn Hiring Rate#
We examined the correlation between the number of conflict events and the LinkedIn Hiring Rate (YOY) across countries. Overall, there is no consistent pattern — most countries show weak or no correlation.
Country-Level Pearson Correlation Coefficients
Country |
Pearson r |
Interpretation |
|---|---|---|
Algeria |
+0.17 |
Very weak positive correlation |
Bahrain |
-0.05 |
Very weak negative correlation |
Egypt |
-0.08 |
Very weak negative correlation |
Iraq |
+0.27 |
Weak positive correlation |
Jordan |
-0.12 |
Very weak negative correlation |
Malta |
+0.16 |
Very weak positive correlation |
Morocco |
-0.54 |
Moderate negative correlation |
Pakistan |
+0.07 |
Very weak positive correlation |
Saudi Arabia |
+0.32 |
Weak positive correlation |
Tunisia |
+0.03 |
Very weak positive correlation |
Qatar |
-0.44 |
Moderate negative correlation |
United Arab Emirates |
+0.91 |
Very strong positive correlation (⚠️ few data points) |
With the exception of Morocco, Qatar, and the UAE, most countries show only weak or inconsistent relationships. The UAE’s very high correlation should be interpreted cautiously due to a small number of observations.
The same pattern was observed excluding event types “Protests” and “Strategic developments”
Regression analysis#
================================================================================================================================================
Model 1: Fatalities Only Model 2: Events Only Model 3: Fatalities + Events Model 4: + Fixed Effects
------------------------------------------------------------------------------------------------------------------------------------------------
Intercept -0.0318*** -0.0324*** -0.0315*** 0.3560***
(0.0101) (0.0109) (0.0110) (0.0360)
total_fatalities 0.0001 0.0001 0.0000
(0.0001) (0.0002) (0.0002)
nr_events 0.0000 -0.0000 0.0001
(0.0000) (0.0001) (0.0001)
C(country)[T.Bahrain] -0.0710***
(0.0257)
C(country)[T.Egypt] -0.0556**
(0.0255)
C(country)[T.Iraq] -0.1156*
(0.0656)
C(country)[T.Jordan] -0.0508**
(0.0255)
C(country)[T.Malta] -0.1276***
(0.0255)
C(country)[T.Morocco] -0.0808***
(0.0304)
C(country)[T.Pakistan] -0.1353*
(0.0771)
C(country)[T.Qatar] -0.0331
(0.0361)
C(country)[T.Saudi Arabia] -0.0621**
(0.0302)
C(country)[T.Tunisia] -0.1178***
(0.0256)
C(country)[T.United Arab Emirates] -0.0501
(0.0390)
C(month)[T.Timestamp('2022-07-01 00:00:00')] -0.0383
(0.0441)
C(month)[T.Timestamp('2022-08-01 00:00:00')] -0.0769*
(0.0432)
C(month)[T.Timestamp('2022-09-01 00:00:00')] -0.1208***
(0.0449)
C(month)[T.Timestamp('2022-10-01 00:00:00')] -0.1332***
(0.0440)
C(month)[T.Timestamp('2022-11-01 00:00:00')] -0.2906***
(0.0451)
C(month)[T.Timestamp('2022-12-01 00:00:00')] -0.3895***
(0.0443)
C(month)[T.Timestamp('2023-01-01 00:00:00')] -0.2993***
(0.0451)
C(month)[T.Timestamp('2023-02-01 00:00:00')] -0.3235***
(0.0448)
C(month)[T.Timestamp('2023-03-01 00:00:00')] -0.4000***
(0.0441)
C(month)[T.Timestamp('2023-04-01 00:00:00')] -0.4660***
(0.0441)
C(month)[T.Timestamp('2023-05-01 00:00:00')] -0.2489***
(0.0462)
C(month)[T.Timestamp('2023-06-01 00:00:00')] -0.4549***
(0.0448)
C(month)[T.Timestamp('2023-07-01 00:00:00')] -0.2966***
(0.0448)
C(month)[T.Timestamp('2023-08-01 00:00:00')] -0.3843***
(0.0449)
C(month)[T.Timestamp('2023-09-01 00:00:00')] -0.4039***
(0.0448)
C(month)[T.Timestamp('2023-10-01 00:00:00')] -0.4523***
(0.0440)
C(month)[T.Timestamp('2023-11-01 00:00:00')] -0.4057***
(0.0440)
C(month)[T.Timestamp('2023-12-01 00:00:00')] -0.3453***
(0.0432)
C(month)[T.Timestamp('2024-01-01 00:00:00')] -0.4207***
(0.0441)
C(month)[T.Timestamp('2024-02-01 00:00:00')] -0.3981***
(0.0450)
C(month)[T.Timestamp('2024-03-01 00:00:00')] -0.5011***
(0.0465)
C(month)[T.Timestamp('2024-04-01 00:00:00')] -0.3565***
(0.0463)
C(month)[T.Timestamp('2024-05-01 00:00:00')] -0.4670***
(0.0465)
C(month)[T.Timestamp('2024-06-01 00:00:00')] -0.4333***
(0.0463)
C(month)[T.Timestamp('2024-07-01 00:00:00')] -0.2989***
(0.0444)
C(month)[T.Timestamp('2024-08-01 00:00:00')] -0.3887***
(0.0451)
C(month)[T.Timestamp('2024-09-01 00:00:00')] -0.3810***
(0.0466)
C(month)[T.Timestamp('2024-10-01 00:00:00')] -0.3272***
(0.0468)
C(month)[T.Timestamp('2024-11-01 00:00:00')] -0.3585***
(0.0455)
C(month)[T.Timestamp('2024-12-01 00:00:00')] -0.3186***
(0.0450)
R-squared 0.0035 0.0022 0.0035 0.6728
R-squared Adj. 0.0003 -0.0009 -0.0029 0.6213
Country FE No No No Yes
Month FE No No No Yes
N 317 317 317 317
R-squared 0.003 0.002 0.003 0.673
================================================================================================================================================
Standard errors in parentheses.
* p<.1, ** p<.05, ***p<.01
Summary of Regression Results#
We estimate the relationship between conflict intensity and LinkedIn Hiring Rate (LHR) across countries and months using four OLS models:
Model 1 includes only total fatalities.
Model 2 includes only number of events.
Model 3 includes both.
Model 4 adds country and month fixed effects.
Key takeaways:
Models 1–3 show very low explanatory power (R² < 0.01), suggesting that fatalities or number of events alone do not meaningfully explain hiring rate changes.
Model 4, which includes fixed effects, improves the model fit substantially (R² = 0.627), highlighting the importance of controlling for country-specific and time-specific factors.
The coefficient for number of events becomes statistically significant in Model 4, but the effect size remains very small.
Total fatalities are not statistically significant in any specification.
Overall, conflict indicators appear to have limited explanatory power for variations in LHR when not controlling for fixed effects. Most of the variation is captured by country and month effects, not by conflict itself.