Jordan: Selected Issues
Author: International Monetary Fund. Middle East and Central Asia Dept.
Publication Date:
09 Jun 2014
eISBN:
9781498368919
Language:
English
Keywords:
ISCR; CR; IMF staff estimate; labor market; Syria crisis; labor market outcome; conflict level; empirical analysis; Syrian refugee; prior research; Labor markets; Current account
Download PDF (1.7 MB)

This Selected Issues paper investigates the macroeconomic impact of the Syria crisis on Jordan. It is indicated that the crisis: (1) had an overall negative impact on measured output growth—although anecdotal evidence suggests possibly a positive impact on output in the informal sector; (2) contributed to inflationary pressures, particularly on rents; and (3) strained labor markets, mostly in the informal sector as refugees compete with locals for jobs. Although the crisis has put a strain on the external trade balance, the overall impact on the current account is not clear.

Abstract

This Selected Issues paper investigates the macroeconomic impact of the Syria crisis on Jordan. It is indicated that the crisis: (1) had an overall negative impact on measured output growth—although anecdotal evidence suggests possibly a positive impact on output in the informal sector; (2) contributed to inflationary pressures, particularly on rents; and (3) strained labor markets, mostly in the informal sector as refugees compete with locals for jobs. Although the crisis has put a strain on the external trade balance, the overall impact on the current account is not clear.

The Syria Crisis as Felt in Jordan1

This note investigates the macroeconomic impact of the Syria crisis on Jordan. We find that the crisis: (1) had an overall negative impact on measured output growth—although anecdotal evidence suggests possibly a positive impact on output in the informal sector; (2) contributed to inflationary pressures, particularly on rents; and (3) strained labor markets, mostly in the informal sector as refugees compete with locals for jobs. While the crisis has put strain on the external trade balance, the overall impact on the current account is not clear.2

A. Output

1. The Syria crisis has weighed on measured growth. On the one hand, losses of the Syrian export routes as well as a hesitant investor sentiment—resulting from uncertainty surrounding the length and outcome of the Syria conflict—are likely to have adversely impacted output. On the other hand, increased consumption brought about by a sizable influx of Syrian refugees has contributed positively to economic activity. Staff empirically assesses the overall impact to be negative, with losses to output growth estimated at about one percentage point in 2013—one third of the impact on Lebanon’s growth, as estimated by a recent World Bank study (see Appendix I for the empirical model).3 In other words, had the Syria crisis never occurred, growth in 2013 would have been at about 4 percent. This is in line with market analysts’ forecast of Jordan’s growth prior to the time when they perceived the Syria crisis as significantly affecting Jordan (Figure 1).

Figure 1.
Figure 1.

Growth Forecasts for 2013

(In percent; Analysts’ forecasts are EIU and S&P average)

Citation: IMF Staff Country Reports 2014, 153; 10.5089/9781498368919.002.A001

Sources: Economist Intelligence Unit; Standard and Poor’s; and IMF staff estimates.

2. Nonetheless, informal sector output is likely to have risen. Even before the Syria crisis, the informal sector was large, with staff’s prior research putting its size at about 26 percent of measured GDP.4 Now, anecdotal evidence suggests an even growing share of unmeasured output due to the heavy involvement of Syrian refugees in informal activity in sectors such as agriculture, construction, food services, retail trade, and home-based production. However, some crowding out of Jordanian nationals may have occurred, as evidenced by the recent sharp decline in Jordanian self employment, a typical proxy for informal sector employment (Figure 2).5

Figure 2.
Figure 2.

Jordanian Informal Sector Workers and Syrian Refugees

(For self employment, 4-quarter moving average of seasonally adjusted data)

Citation: IMF Staff Country Reports 2014, 153; 10.5089/9781498368919.002.A001

Sources: Jordanian authorities; United Nations High Commissioner for Refugees; and IMF staff estimates

B. Labor Markets

3. At the national level, statistical evidence does not appear to support the hypothesis that the Syria crisis had negatively impacted formal labor market outcomes. Unemployment among Jordanians has increased in 2013, because of a decline in the ratio of employment to working age population and despite lower labor force participation rates (Figure 3). One could argue that the massive influx of Syrian refugees has something to do with such unfavorable outcomes, for example, through their competition with Jordanians for low-skilled formal jobs. Statistically, and based on available data, staff finds no evidence supporting this hypothesis (see Appendix II for the empirical model). This is supported by the fact that registered refugees, although considered Jordanian residents, are not legally allowed to work in Jordan.6 This constrains them from competing for jobs in both the public sector and the formal private sector, and hence limits their employment options to the informal sector. Such a conclusion is reinforced by the fact that many of the refugees are very young—more than half of the total refugee population is less than 18 years old. At this age, competition with locals would likely be in the informal labor market. Instead, the public sector creating fewer jobs is a plausible alternative explanation of the recent worsening of unemployment outcomes, because the sector has been generating much of the jobs in the last decade (Figure 4).

Figure 3.
Figure 3.

Labor Market Outcomes

(In percent)

Citation: IMF Staff Country Reports 2014, 153; 10.5089/9781498368919.002.A001

Sources: Jordanian authorities; and IMF staff estimates.
Figure 4.
Figure 4.

Unemployment and Public Sector Employment

(In percent; 4-quarter moving average of seasonally adjusted data)

Citation: IMF Staff Country Reports 2014, 153; 10.5089/9781498368919.002.A001

Sources: Jordanian authorities; and IMF staff estimates.

4. The same result holds when one looks at labor market outcomes by governorates. Amman, Mafraq, Irbid, and Zarqa, cumulatively host nearly ninety percent of the Syrian refugees; eight governorates, including those in the south, host the remaining 10 percent. In the former group of governorates, unemployment rates have somewhat declined from their pre-Syria conflict levels, whereas in the latter group, such rates have marginally deteriorated (Figure 5). This is consistent with staff’s empirical analysis.

Figure 5.
Figure 5.

Labor Market Outcomes by Governorates

(In percent)

Citation: IMF Staff Country Reports 2014, 153; 10.5089/9781498368919.002.A001

Sources: Jordanian authorities; and IMF staff estimates.

C. Inflation

5. The influx of Syrian refugees has put upward pressure on the prices of non-tradable goods. In particular, staff finds a statistically significant effect on Jordan’s rental prices last year, with an estimated crisis impact on end-of-period rent inflation of about 5 percentage points by year-end (see Appendix III for the empirical model). This translates into a higher end-of-period consumer price index (CPI) inflation of about 0.6 percentage points relative to the counterfactual of the Syria crisis having not occurred (Figure 6). Moreover, staff finds no statistically significant effect of the crisis on domestic food prices. The latter is estimated to be predominantly driven by international food prices, both in the short run and long run, likely reflecting Jordan’s high degree of openness.

Figure 6.
Figure 6.

Syria Crisis Impact on Inflation, 2013

(In percent)

Citation: IMF Staff Country Reports 2014, 153; 10.5089/9781498368919.002.A001

Source: IMF staff estimates.

D. Current Account

6. The Syria crisis has taken a toll on the trade balance. Because of the conflict, Jordan lost a major export route to Europe and other countries in the region. Indeed, during the first eleven months of 2013, combined exports to Lebanon, Turkey, and Europe dropped by 30 percent relative to the same period in 2012. As a result, Jordan is reportedly considering more expensive alternative routes—including Iraq and the Aqaba Port—with adverse implications for competitiveness. Abstracting from mining exports, which have been facing difficulties unrelated to the Syria conflict, the severe contraction in exports that travel through Syria has resulted in a sizable moderation in export growth—excluding potash and phosphate, exports grew by an average of a mere 0.7 percent per quarter in the two years up to 2013-Q3, down from an average of 3 percent in the previous two years (Figure 7). At the same time, imports through Syria have declined sharply (Figure 8). Nonetheless, the increased size of the population brought about by the influx of Syrian refugees has resulted in a rise in overall imports. Non-energy imports are estimated to have increased by over 11 percent in 2013.

Figure 7.
Figure 7.

Exports

(4-quarter moving average of seasonally adjusted data; in logs)

Citation: IMF Staff Country Reports 2014, 153; 10.5089/9781498368919.002.A001

Sources: Jordanian authorities; and IMF staff estimates.
Figure 8.
Figure 8.

Imports

(4-quarter moving average of seasonally adjusted data; in logs)

Citation: IMF Staff Country Reports 2014, 153; 10.5089/9781498368919.002.A001

Sources: Jordanian authorities; and IMF staff estimates.

7. Nonetheless, the overall impact on the current account is not clear. Transfers through the UN agencies and from other private citizens abroad to support the refugees have contributed positively to the current account; but there is uncertainty regarding their size, particularly on the latter. Tourism got hit initially when the transition started in the region. Because Jordan has been stable, tourism has begun to recover, including from non-Arab countries (Figure 9). That said, recent trends in tourism receipts are likely to be underestimated because the data do not take into account increased duration of stay of Gulf Cooperation Council (GCC) tourists as a result of reorientation of such activities towards Jordan from more affected countries—including Lebanon, Egypt, and Syria. This may apply as well to tourists from other countries in the region.7 The overall impact of the crisis on the current account though is not clear, because it depends on whether the gains from current private transfers and tourism receipts more than offset the deterioration in the trade balance. Available data do not allow for a quantitative assessment.

Figure 9.
Figure 9.

Tourism Receipts

(4-quarter moving average of seasonally adjusted data; in logs)

Citation: IMF Staff Country Reports 2014, 153; 10.5089/9781498368919.002.A001

Sources: Jordanian authorities; and IMF staff estimates.

Appendix I. The Impact on Measured GDP Growth—Empirical Model

1. We begin by estimating the following dynamic error correction model utilizing quarterly data:

Δ⁢yt=δ0⁢Dt+Σi=1pαiΔ⁢yt−i+β(y−y*)t−1+λ⁢ syriat+ɛt

where yt denotes the log of seasonally-adjusted real GDP; Δy is first difference of yt, measuring the growth rate of real output; y-y* denotes the output gap measured as the deviation of the log of seasonally-adjusted real GDP from its Hodrick-Prescott (HP) trend; Dt is a vector of deterministic variables that includes a constant and a time trend—the latter accounts for the combined effect of potentially omitted variables. Conceptually, one can measure the Syria shock by two alternative ways: the number of refugees or a dummy variable. The lack of sufficient time series data on the number of refugees renders the first approach not feasible. As such, we conduct our empirical analysis using a dummy variable to proxy for the Syria crisis, denoted Syria above. A priori, we expect ß to be negative (signifying the workings of error correction). Also, a negative value for λ would indicate a negative overall impact on GDP growth of the Syria crisis.

2. Pretesting the above general model and starting a with relatively generous lag length (p=6 in the above model) revealed the following: Spillovers of the Syria crisis are not statistically significant prior to the second quarter of 2012; only the fifth lag on growth matters statistically; large estimated errors in 1993-Q4 and 1995-Q1 taint residual normality and homoscedasticity; and the estimated coefficient of the time trend is not statistically significant. Utilizing this information, we estimate the following final parsimonious error correction model:

Δ⁢yt=0.009−0.042⁢ Dt93,Q4+0.040 Dt95,Q1+0.288Δyt−5−0.540(y−y*)t−1−0.009⁢ syriat(0.000)(0.0001)(0.0004)(0.0017)(0.0000)(0.0452)+ɛt

where Dt93,Q4 and Dt93,Q4 are impulse dummy variables taking the value of 1 in 1993-Q4 and 1995-Q1 respectively and zero otherwise; and Syriat is a dummy variable that takes the value of zero prior to 2012-Q2 and one otherwise. The regression output shows that all the estimated parameteres are highly statistically significant (p-values of the t-statistic are reported in brackets); the Syria crisis has weighed on Jordan’s output growth; and there is ultimate convergence to the steady state, as indicated by the negative coefficient on the lagged output gap. Furthermore, various diagnostis tests show that such inference comes from a statistically well specified and constant model with white noise residuals.1 The coefficient of the Syria dummy variable together with the estimated model dynamics indicate losses to GDP growth (because of the Syria crisis) of one percentage point in 2013.

Appendix II. The Impact on Formal Labor Market Outcomes—Empirical Model

1. We examine the impact of the Syria crisis on labor market outcomes by estimating dynamic Okun’s-type regressions augmented to account for the Syria shock. We estimate these models at both the national level and for the those governorates that host ninety percent of the refugees. Specifically, and utilizing seasonally-adjusted quarterly data, we estimate the following model for each of the three main labor market outcomes (LMO): the unemployment rate, the employment-to-working age population ratio, and the labor force participation rate.

LMOt=α0+Σi=1pαiLMOt−i+Σi=0pβtgapt−i+λ⁢ Syriat+ɛt

where LMO denotes a given labor market outcome (in percent); and gap denotes the output gap, measured as the percent deviation of the log of real GDP from its HP trend. As in the previous section, we model the Syria shock through the use of dummy variables. The lag-length of a given model is not known a priori, but determined through sequential F-tests—a general-to-specific modeling approach.

2. Table A2.1 shows a summary of the regression results at the national level. The top block shows results for unemployment. The first row denoted Syria 12-1 shows results for the model where the Syria dummy variable takes the value of zero prior to 2012-Q1 and one otherwise. It reports the estimated coefficient of the Syria dummy variable (λ), its t-statistic, and the associated p-value. It also reports the p-values of a battery of diagnostic tests for model validity with no rejections of the null hypotheses, underpinning the empirical validity of the model (see description in Appendix I). Table A2.1 shows that the estimated coefficient of the Syria dummy variable is not significant. The remaining rows in the unemployment block report results with alternative timing of potential spillovers materializing from the Syria shock. That is, we experiment with models with Syria 12-2 (equal to zero before 2012-Q2, and one otherwise); Syria 12-3 (equal to zero before 2012-Q3, and one otherwise); etc.. None of the Syria dummy variables are significant, and once again the models are empirically valid. The middle and lower blocks of Table A2.1 report results for labor force participation and employment to population ratios respectively.1

Table A2.1.

Estimated Impact of Syria Crisis on Formal Labor Market Outcomes at the National Level

article image
Source: IMF staff estimates.

3. Table A2.2 repeats the same exercise for those governorates that host ninety percent of the Syrian refugees. The evidence from Table A2.1 and Table A2.2 shows that for all the 42 regressions estimated, the Syria conflict does not appear to have had a statistically significant impact on Jordan’s formal labor market outcomes. And such inference is derived from models that are statistically valid.

Table A2.2.

Estimated Impact of Syria Crisis on Formal Labor Market Outcomes for Governorates Hosting 90 Percent of Syrian Refugees

article image

Source: IMF staff estimates.

Appendix III. The Impact on Inflation—Empirical Model

Food prices

1. We estimate the following Autoregressive Distributed Lag model (ADL) utilizing seasonally-adjusted monthly data:

pt=δ0Dt+Σi=1pαipt−i+Σi=0pϕipt−i*+λ⁢ Syriat+ɛt

where p is the log of the index of Jordanian food prices; p* is the log of the index of international food prices; Syriat is a dummy variable that captures the Syria shock; and Dt is a vector including a constant and a deterministic time trend—the latter to capture the combined effect of all potential variables affecting domestic food prices other than international food prices and the Syria conflict.

2. Unit root tests indicate that both domestic and international food prices are non-stationary or integrated of order one. Given that Jordan is a very open economy, we would expect domestic and international food prices to co-move over time (or share a common stochastic trend). As such, we posit the following co-integration relationship between the two sets of prices.

pt=η0Dt+β1pt*+β2⁢ Syriat+ɛt

where the ßs are interpreted as long-run coefficients. In fact, this model is the long-run solution of the ADL model where

β1=Σi=0pϕi1−Σi=1pαi⁢ and⁢  β2=λ1−Σi=1pαi

3. Table A3.1 presents the estimation results of the ADL model and its long-run solution for a battery of dummy variables each capturing a different potential starting time of Syria’s spillover effect. For example, Syria 12-1 takes the value of zero before 2012-Q1, and one otherwise; Syria 12-2 takes the value of zero before before 2012-Q2, and one otherwise, etc..

Table A3.1.

Estimated Impact of Syria Crisis on Food Prices

article image
Source: IMF staff estimates.

4. The Syria conflict appears not to have a statistically significant effect on the Jordanian food price index—all dummy variables are not significant. Rather, international food prices appear to be the key driver.

Rental prices

5. We empirically estimate an autoregressive process for rent inflation augmented to account for the Syria shock. Statistically, the model captures the true data-generating process remarkably well and the residuals are white noise. Spillover effects from Syria appear to materialize in the second half of 2013 (Table A3.2).

Table A3.2.

Estimated Impact of Syria Crisis on Rents

article image
Source: IMF staff estimates.

Δrt is the first difference of the rental price index. The data are bi-annual and seasonally not adjusted. As such, a seasonal dummy variable is included. Aside from the third lag of the first difference, all lags are not significant. Syria 13-June is a dummy variable that takes the value of zero prior to June 2013 and one otherwise.

1

Prepared by Yasser Abdih.

2

This note does not look into the poverty and fiscal implications of the Syria crisis. Regarding poverty, the lack of time series data substantially hinders any meaningful inference. And as for the fiscal impact, it was examined in detail in a recent USAID study.

3

World Bank, “Lebanon: Economic and Social Impact Assessment of the Syrian Conflict,” Report No. 81098-LB, 2013.

4

Abdih, Yasser, “MENA Oil Importers: Addressing Informality and Promoting Inclusion,” in Regional Economic Outlook, Middle East and Central Asia; International Monetary Fund, October 2011.

5

In Figure 2, we apply a 4-period moving average to remove irregularities and volatility in the data, thereby allowing for a clearer trend. This is done for presentational purposes only, and none of the results hinges on such a transformation. The same observation applies to other figures in this note that use a 4-period moving average.

6

Syrian refugees are not covered in labor market surveys. For documentation on the illegality of granting work permits for Syrian refugees, see United Nations and the Government of Jordan, “Needs Assessment Review of the Impact of the Syrian Crisis on Jordan,” November 2013.

7

Two additional factors may be at play here. First, there is the “Libya effect.” Libyans came to Jordan for medical treatment throughout 2012, but this was not repeated in 2013. This has nothing to do with the Syria crisis, and may partially explain the moderation in tourism receipts in 2013 from other Arab countries (see Figure 9). Second, there is a “Ramadan effect.”Ramadan in 2013 coincided with the tourism season and may have negatively affected tourism from the GCC and other Arab countries. Again, this has nothing to do with the Syria crisis, and although the data in Figure 9 are seasonally-adjusted, the effect may still be there, because Ramadan does not come at the same time each year.

1

Diagnostic tests for model validity (in the format Statistic = value (p-value)) test the null hypotheses that: there is no residual autocorrelation (AR); there is no residual autoregressive conditional heteroscedasticity (ARCH); the residuals are normally distributed; there is no residual heteroscedasticity (Hetero); the estimated parameters are stable over time (Chow); and the estimated regression model is well specified (Reset). Specifically, AR = 1.64 (0.161); ARCH = 1.19 (0.323); Normality = 2.50 (0.286); Hetero = 0.57 (0.726); Chow =0.59 (0.916); and Reset = 1.35 (0.266).

1

Unit root tests suggest that the unemployment rate is stationary while labor force participation and the employment to population ratio are I (1). As such, the first difference of the latter two is used in the empirical specifications.

Same Series

Other IMF Content

Other Publishers

Cover IMF Staff Country Reports
Jordan: Selected Issues
Author: International Monetary Fund. Middle East and Central Asia Dept.
Volume/Issue:
Volume 2014: Issue 153
Publisher:
International Monetary Fund
ISBN:
9781498368919
ISSN:
1934-7685
Pages:
28
DOI:
https://doi.org/10.5089/9781498368919.002
  • View in gallery

    Growth Forecasts for 2013

    (In percent; Analysts’ forecasts are EIU and S&P average)

  • View in gallery

    Jordanian Informal Sector Workers and Syrian Refugees

    (For self employment, 4-quarter moving average of seasonally adjusted data)

  • View in gallery

    Labor Market Outcomes

    (In percent)

  • View in gallery

    Unemployment and Public Sector Employment

    (In percent; 4-quarter moving average of seasonally adjusted data)

  • View in gallery

    Labor Market Outcomes by Governorates

    (In percent)

  • View in gallery

    Syria Crisis Impact on Inflation, 2013

    (In percent)

  • View in gallery

    Exports

    (4-quarter moving average of seasonally adjusted data; in logs)

  • View in gallery

    Imports

    (4-quarter moving average of seasonally adjusted data; in logs)

  • View in gallery

    Tourism Receipts

    (4-quarter moving average of seasonally adjusted data; in logs)

Growth Forecasts for 2013

(In percent; Analysts’ forecasts are EIU and S&P average)

Jordanian Informal Sector Workers and Syrian Refugees

(For self employment, 4-quarter moving average of seasonally adjusted data)

Labor Market Outcomes

(In percent)

Unemployment and Public Sector Employment

(In percent; 4-quarter moving average of seasonally adjusted data)

Labor Market Outcomes by Governorates

(In percent)

Syria Crisis Impact on Inflation, 2013

(In percent)

Exports

(4-quarter moving average of seasonally adjusted data; in logs)

Imports

(4-quarter moving average of seasonally adjusted data; in logs)

Tourism Receipts

(4-quarter moving average of seasonally adjusted data; in logs)