The Role of Political Connections in COVID Policy Response: Effectiveness of Firm-Level Government Support in Egypt

Abstract The COVID-19 pandemic saw two sets of policy responses: lockdown to limit spread of the virus, which was a huge demand and supply shock, and government support to firms and individuals to offset the effects of this policy-induced shock. This paper explores the allocation and effectiveness of government support to firms in Egypt. We consider both financial support measures which were by and large already being implemented pre-COVID, as well as tax- and loan-related exemptions and deferments. After controlling for the endogeneity of government support, our main findings show that the latter has helped mitigate the effects of COVID-19, with a significantly larger, favorable impact on smaller, younger and private firms. There is no equity-effectiveness trade-off. However, although these firms apparently make better use of government support, they receive a disproportionately smaller share of it. In line with the emerging ‘unsocial’ social contract, government support has been chiefly determined by political connections and a captured industrial policy. This ‘misallocation’ reinforces the ‘missing middle’ phenomenon which acts as a constraint as SMEs are unable to grow. Nevertheless, the crisis has presented a chance for the pattern of support to slowly shift towards the more vulnerable through the more frequent use of ‘exemptions and deferments’.


Introduction
The pandemic has had severe economic consequences around the world. Most countries have instituted full or partial lock-down measures to save lives during the pandemic. To mitigate the spread of the virus the Egyptian government introduced a partial lockdown as of March 2020.
The lockdown restricted opening hours and movement with the exception of grocery shops and supermarkets (El-Tawil 2020). Further restrictions were introduced in April 2020 which lasted until the end of May 2020. Another round of less strictly-enforced restrictions were introduced during the second wave of COVID-19 from 1 December 2020 to January 2021.
However, protecting human life has an economic cost. Lockdown measures led to both demand and supply-side shocks which resulted in a global economic crisis. Substantial numbers of businesses have been forced to exit the market or to temporarily close. Global supply chains have been substantially disrupted resulting in increased supply costs limiting, in turn, firms' ability to enforce the quality and timeliness of contracts (Demertzis &Masllorens, 2020 andAyadi et al. 2022).
In Egypt, despite the fact that there was positive GDP growth in 2021, manufacturing gross value added declined by nearly 6%. Construction and agriculture were the source of GDP growth during the pandemic (Ministry of Planning & Economic Development, 2022). Consequently, numerous firms have shut down, suffered lower productivity, lost previous productivity gains and have seen their sales and profits shrink (Bloom et al. 2020). Surviving firms have adopted various strategies to cope with the pandemic such as reducing input costs through worker layoffs and salary adjustments, for example.
To mitigate the shock on both firms and households the government of Egypt rolled out a fullfledged stimulus package. This included a number of fiscal and monetary measures. On the fiscal front, the government announced a $6.13 billion packageequivalent to 1.8% of GDP (Krafft et. al. 2021). To support government revenues, a Corona tax of 1% has been levied on all public and private sector salaries and of 0.5% on state pensions. These revenues were supposed to fund fiscal measures to support negatively affected sectors, namely cutting taxes on dividends and real estate, fast-tracking payouts from the Export Subsidy Fund, expanding the Social Security and Pension Act's coverage; providing one-time stipends of EGP 500 for seasonal and temporary workers; and postponing the filing deadline for auditors and SMEs (El-Haddad, 2020a).
With respect to monetary and financial policy, a preferential interest rate (8%) has been set for the loans of some industries such as tourism, manufacturing, agriculture and construction sectors, and on mortgages for low-income and middle-class housing. The aim of these measurers was to counter the contractionary effects of the pandemic through encouraging industrial sector growth and capital expenditure lending. Furthermore, the Central Bank of Egypt (CBE) has provided short-term loans to micro-enterprises and small and medium enterprises (SMEs) of up to a year to cover their operational expenses. In addition, the share of bank loan portfolios that must be allocated to SMEs has been raised from 20 to 25 percent. The Financial Regulatory Authority (FRA) also announced a delay of up to 50% of the value of monthly installments for micro-borrowers. Our data show that postponement of loan installments repayment, delay of loan service payments, reductions and discounts on given loans and tax payment deferments have been the most frequently used government support measures in the manufacturing sector in the country.
There are two important and interdependent considerations in thinking about government support. The first relates to vulnerability and equity: whether to target sectors and/or firm types (e.g. micro and SMEs) most affected by the crisis, or to give support to those sectors or firms whose performance is most likely to be significantly improved with limited support. The second approach implies giving support to those least in need. This is similar to considering targeting on poverty headcount versus the depth of poverty, since it is much easier to move people just under the poverty line to cross it in order to significantly improve the indicator. Another analogy is the triage argument in medicine. When resources are limited those with the highest likelihood of survival should be supported and not those with the deepest wounds.
The second consideration is the effectiveness of the actual support received by sectors or firms. Did support prevent firm closures, the laying off of workers or large reductions in revenues or profits? From an economic perspective, support should go where it is most effective.
It may be the case that government support is most effective when given to the most vulnerable. In this case, there is no equity-effectiveness trade-off. Whether this is the case, is the core of what this paper investigates.
In Egypt, as in other countries with poorer governance, support can be shaped by the prevalent state-business relationships that are mediated by the excessive degree of capture of industrial policy. In Egypt, the 'unsocial' social contract emerging under liberalization meant that the state used trade, industrial and other economic policies to favour an emerging group of crony capitalists who in turn provided support for the regime (El-Haddad, 2020b). Thus, it is important to look into whether the COVID-19 crisis has induced a shift in the entrenched pattern of support.
In this paper, we analyze a new data set from our 2020/21 Egyptian Industrial Firm Behaviour Survey (EIFBS) to first look into which types of firms have received government support, both pre-and post-pandemic and the underlying determinants of the existing distribution. We then assess the effectiveness of this support in curbing the negative impacts of the shock. We focus on six main performance indicators: employment, revenues, losses in profits, reduction in working hours, layoffs and whether the firm has ever closed down since the start of the pandemic. The latter four measures capture more the immediate impacts of pandemic induced short-term changes, whereas revenue and employment may be regarded as longer run effects.
Post-COVID government support was received by a quarter of the surveyed firms and is divided into two broad categories: (1) financial support; and (2) exemptions and deferments. The former are further divided into: (1) financial and technical support towards factors of production; and (2) general and other financial support. While we cover all categories with the range of all 14 underlying government support measures, we pay particular attention to the top five used measures post-COVID. These all fall under the 'exemptions and deferments' category, which captures more the short-term support measures designed to swiftly mitigate the damaging effects of the crisis, which were received by about 84% of all firms receiving post-COVID support (see also Krafft et al. 2021 on types of firm support in Egypt). In contrast 'financial support' has generally been of a longer-term nature, already being received by many firms pre-pandemic.
Evaluating effectiveness of government support is a crucial question to inform the efficient use of limited government resources, improve budget allocations and to identify effective measures to strengthen firm resilience in the face of future shocks. The empirical literature assessing the effectiveness of these programs and especially on the region is still relatively scarce, this paper fills this gap. Bennedsen, Larsen, Schmutte, and Scur (2020), use survey data for small, medium and large firms. They compare firms' actual layoff and furlough decisions to the reported counterfactual decisions in the absence of government aid. Estimating that 81,000 fewer workers were laid off and 285,000 fewer furloughed they conclude that government support had been effective in preserving jobs. Similarly, Lalinsky and P al (2021), using data for Slovakian firms, show that government support has helped save jobs and sustain economic activity. This is different for high productivity, privately owned and exporting firms in Portugal, which generally did not reduce their employment levels despite not having received government support (Kozeniauskas, Moreira, & Santos, 2020).
Our main findings show that government support has helped mitigate the effects of COVID-19, with a significantly larger, favorable impact on smaller, younger and private firms. Thus, government support is most effective when given to the most vulnerable. However, although these firms apparently make better use of government support, they receive a disproportionately smaller share of it, which reinforces the 'missing middle' phenomenon, which is further reinforced by political connections and a persistent soft budget constraint faced by public sector firms.
The 'missing' middle phenomenon refers to the fact that the distribution of firms is such as there is a concentration of a number of very old, mostly large, firms on the one hand and very large numbers of very small, younger firms on the other ( This phenomenon is particularly marked in the MENA region, where micro, medium and small sized enterprises (MSMEs) in MENA's oil importing countries (OICs) are about 10 years older than either their East Asia and Pacific or their Europe and Central Asia comparators (El-Haddad, Adel, Abdel-Latif, & Terefe, 2021). Egypt has the oldest firms of all OICs.
In line with the 'unsocial' Social Contract (El-Haddad, 2020b), government support has been chiefly determined by political connections and a captured industrial policy. Political connections are perpetuated by the fact that post-COVID support is, in large part, determined by pre-COVID government support. A large chunk of firms which received support pre-COVID also received it post-COVID. There is a practical aspect to such an approach since systems of support were already in place pre-COVID, making it straightforward to identify firms which had previously received support, compared to identifying others that are unknown and most likely not politically connected to government. Nonetheless, the fact that political connections largely influence the allocation of support in response to the pandemic may mean that it is not allocated in the most efficient manner.
In contrast, compared to 'financial support', post-COVID support on 'exemptions and deferments' had been relatively more allocated to favor smaller, younger and private firms where it is more effective. Hence, there is some evidence that the COVID-19 crisis has induced some shift in the entrenched pattern of support. This short-term, relatively more crisis induced category of support has reduced post-COVID firm layoffs which is not the case for the pre-existing 'financial support' interventions. More nuances arise when looking at the six measures most frequently implemented post-COVID: export drawback, loan and tax related measures are the most effective. It has indeed been found that tax relief and support through the formal banking channel may not reach most firms in developing countries, but it can keep otherwise viable firms from slipping into informality (Mora, 2020). With respect to the least frequently used financial support, purchase and lease of industrial land is the most effective.
The following section first describes the EIFBS data followed by a depiction of the stylized facts and the 'missing middle'. Next, is a section on the methodology. Section four presents the empirical findings and section five proceeds with a discussion of results and concludes.
2. EIFBS survey instrument, stylized facts and the 'missing middle'

EIFBS survey instrument and sampling design
We use unique and recently collected data from the self-designed 2020/21 Egyptian Industrial Firm Behavior Survey (EIFBS) of 2,383 Egyptian manufacturing firms. The data were collected at the beginning of the second wave of COVID-19 extending to the height of it. 1 More detailed information on sampling is can be found in Supplementary Annex 1.
Two questionnaires were administered, one for firms that are still in operation, and another, very similar one 2 , for firms that have exited the market or have temporarily shut down operations. The response rate is 75%, meaning that we successfully interviewed 2,383 establishments of which 2,338 are in operation and 45 firms that have either exited the market or are temporarily closed. Of the 766 firms we could not interview, an unknown number, and presumably a much higher proportion, have also exited the market. 3 The questionnaire includes 14 modules: basic firm identification data, firm size, firm expectations on recovery and potential exit, changes in firm performance, pandemic transmission channels, ownership and management characteristics, innovation, management practices and use of information technology (IT), production costs, obstacles to operation, exports and global value chains, obstacles to exports, worker training and government support.
As per government support, our analysis focuses on all types of government support before and after the pandemic. We distinguish between financial and non-financial support. Financial support is handed to the firm in the form of direct monetary payments. Other types of support fall under exemptions or deferment of due payments by the firm to government or banking institutions. Exemptions and deferments capture more the short term nature of support measures designed to swiftly dim the damaging effects of the crisis on employment or revenue for example.
In detail these two broad categories can be divided in three subcategories. We divide 'financial support' into two subcategories. First is financial and technical support towards factors of production for: (1) the purchase or lease of land; (2) workers' insurance payments; (3) preparing tenders, auctions or bids; and (4) for the production process. The second sub-category is general and other financial support for: (1) feasibility studies; (2) legal fees; (3) general financial support; (4) swift repayment of old state dues; and (5) refund of export burdens (export drawback). Exemptions and deferments cover: (1) postponement of repayment of bank loan installments; (2) reduction or discount on bank loans; (3) deferral of loan service payments after the due date; (4) deferral of tax payments (income/sales) and (5) exemptions or reductions of tax payments (income/sales).

Stylized facts
Numbers of firms receiving any form of government support has increased by just under a third, increasing from a total of 383 firms pre-COVID to 506 post. Still a modest share presenting about a quarter of all surveyed firms. The frequency of implemented support measures has increased by 17% from 886 to 1035. Figure 1 shows that there has been notable shifts in the types of support measures in response to COVID-19. Three categories have nearly doubled in terms of proportion of use. They fall under the 'exemptions and deferments' support category: postponement of loan repayments (42.3% of all firms receiving support compared to just 25% prior to COVID); tax payment deferral (9.7% compared to 3.5%) and tax exemptions or reductions (5.3% compared to 1.9%). These three in addition to the other two measures of this category, namely deferral in loan service payments (16.5%) and reductions and discounts on loans (10.5%), represent the most measures implemented post-COVID, precisely 84% of all firms receiving support have received any of these 5 measures. In comparison, together with the three loan-related measures, export drawback and financial support with purchasing or renting of land represented the top five administered government support measures prior to the pandemic.

COVID-support and firm characteristics
There is a dominance of large firms receiving post COVID-19 financial support ( Figure 2), both in support of factors of production, general and other types (e.g. for feasibility studies or legal fees). In contrast, SMEs have received above average 'exemptions and deferments; such as postponements to repay loans or their service, reductions in taxes and deferral of payments. Nevertheless, consistent with the literature (El-Haddad et al., 2021;Freund, 2021), larger firms have still received more than proportional post COVID 'exemptions and deferments' support. Table S1 in the supplementary material shows that in our sample, SMEs represent 83% in the sample, yet only 13.5% (weighted) of all SMEs receive this support category. This is in sharp contrast to large firms. While they represent less than 17% of all firms, a higher proportion of 18.2% (of all large firms) receive such support compared to their SMEs counterparts.
Similarly, older firms 4with a mean age above the median of 20 yearshave disproportionately received greater 'financial support' compared to younger firms 5 , particularly the support directed at factors of production ( Figure 3). The majority of the 'exemptions and deferments' goes to the younger firms, in line with the announced purpose of this category's measures.
The role of political connections in COVID policy response 1217 The share of young firms receiving this support category is double the share of old firms that have received it. 6 In terms of firm ownership, the split for receipt of financial support for factors of production is more or less equal, with the exception of public firms receiving a higher proportion for land purchase or lease (59%) and for feasibility studies (73%, Figure 4). Another exception is export drawback which goes predominantly to private sector firms. As intended by the short term debt and tax support measures, in absolute terms, private firms have indeed received the majority of deferments than have their public counterparts. However, in relative terms, 14.4% of all private firms were recipients of this support category, only marginally higher than the share of their public sector counterparts (12%). Nevertheless, given that private firms account for 96% of all firms in the sample (Table S1) this distribution of support continues to be extremely skewed towards public sector firms.

COVID-support, role of the history of support and of political connections
Included among our potential determinants of support receipt is the history of received support as well as political connections known to be binding in the majority of MENA countries. Figure 5 indicates that firms that have been receiving government support in the past, especially that of a financial and longer term nature, are very likely to be the ones also receiving the same type of support post COVID-19 ( Figure 5). Indeed, only 0.5% of all firms have received 'financial support' post-COVID while not having received it prior to COVID and in fact 31% of all post-COVID support was continuing support that was provided both before and after COVID (Table S2). Old structures of support are already in place which makes identifying those firms straightforward compared to others. The same is not strictly true with additional support provided under 'exemptions and deferments' designed to directly target the effects of the crisis and to Source: Authors' own elaboration using the EIFBS. Note: FS stands for financial support. The role of political connections in COVID policy response 1219 potentially reach new segments of firms. In fact, over 10% of firms were newly targeted with measures of that category post-COVID. A firm is said to be politically connected if it has or ever had a government official among its owners, managers or board of directors. On average a larger proportion of politically connected firms are receiving post COVID-19 financial support with the exception of trivial support such as with legal fees and tenders ( Figure 6). Export drawback support hardly goes to politically connected firms (PCF). PCFs are unlikely to be exporters in the first placeas are public firms Source: Authors' own elaboration using the EIFBS. -since they largely benefit from the captive domestic market. Non-politically connected firms receive the bulk of short term debt and tax related facilitations. Table 1 presents a cross-tabulation of the mean value of each performance indicator against the various government support measures. The data show that performance is positively associated with support. Levels of employment and revenues are greater, layoffs and reductions in profits and working hours are less post COVID-19 compared to their non-benefitting counterparts, as is being less likely to have ever closed since begin of the pandemic. 7 2.5.1. The 'missing middle'. Recent literature has referred to the 'missing middle' phenomenon in the Middle East and North Africa Region (MENA) which refers to the fact that the distribution of firms is such as there is a concentration of a number of very old, mostly large, firms on the one hand and very large numbers of very small, younger firms on the other 8 (Schiffbauer et al., 2014, Diwan et al., 2016, Rijkers et al., 2014, El-Haddad, 2020b). This phenomenon is very marked in the MENA region. With an average age of 21 years, MSMEs in MENA oil importing countries are about 10 years older than either their East Asia and Pacific or their Europe and Central Asia comparators (El-Haddad et al., 2021, El-Haddad & Zaki, 2022a. Egypt has the oldest firms (23 years) on average followed by Tunisia and Morocco (20). Jordan and Djibouti have younger firms, but still quite old at 16 9 years. Larger firms are older, with age ranging from 18 for small, 23 for medium (between 20 and 100 workers), 26 for large (>100 workers) and 36 years of age for extra-large (>600 worker) firms. The latter group is twice as old as the group of small firms in our sample.

Government support and firm performance
Given this distribution, it is important to identify (1) whether the pattern of support is more skewed towards politically connected, older, larger and/or public firms 10 ; and (2) if government support provided to this group is more or less effective compared to their smaller, younger and private counterparts.

Methodology
In order to examine the determinants of government support and its effect on firms' performance, we proceed in two stages through an instrumental variable approach to control for the endogeneity between firms' performance and post-COVID government support. Source: Authors' own elaboration using the EIFBS. Note: Yes (No) indicates firms that are (not) politically connected.

Source:
Authors' own elaboration using the EIFBS.

11
(1) Where Gov.SupAC ijk is a dummy variable that takes the value of 1 if firm i in sector j in governorate k benefited from any post-COVID support program and zero otherwise. Z is a vector of control variables including the firm's age 12 and size 13 , whether it is privately or publicly owned, is formal, is located in an industrial zone and whether it is exporting. rj and ck are sector and governorate dummies respectively to control for unobservables at the sector and governorate levels. is the error term. As shown in the stylized facts section above, two variables are expected to affect the likelihood of access to post-COVID government support and are thus used as instruments, to control for endogeneity between the performance variables and government support post-COVID. These two variables are whether the firm is politically connected (Pol.Con.) and whether it had access to any government support prior to COVID (Gov.SupBC). Despite covering plenty of measures, the detailed government support measures do not cover all possible types of government support. This is why we add political connections as instrument since they could influence a firm receiving government contracts, for instance.
A good instrument is one that does not exert a 'direct' influence on outcomes, but does so solely through another explanatory variable -here through post-COVID government support (Gov.SupAC ijk ). Thus, to guarantee the validity of our chosen instruments we ran regressions, where the dependent variables are the performance indicators and the regressors are identical to the variables given in Equation (1) above. The results are insignificant in almost all regressions confirming the results of the Sargan and Basman test of overidentifying restrictions for our instruments and their validity. 14 This implies that while political connections can improve a firm's access to government support or contracts, access in itself does not guarantee that the latter will be efficiently used to improve firm performance. This is in line with Abou-Shady and Zaki (2019) who find that, while state owned firms seem to enjoy the privilege of entering the exports market (i.e. at the extensive margin) due to relaxed barriers to entry, privileged access to information, and formal and informal communication channels with the authorities, they arehoweverunable to compete later on to expand their export activity (i.e. at the intensive margin).
We add an additional instrument of post-COVID government support, a shift share instrument (Gov. Sup.AC Sec.Reg), measured as the share of firms in the same industry and in the same governorate -less the firm in question -that have received post-COVID government support for the respective category. The rationale behind this instrument is that in the same agglomeration (measured by the industry in a specific region), the government may have an incentive to give support to similar firms as they may be aware of the support received by neighboring firms. By removing the firm in question we reduce the likelihood for endogeneity of government support received in simultaneous periods (cf. Edgar S. Dunn 1960in Chun-Yun & Yang, 2008. Again, this instrument follows the principle that a valid instrument induces changes in the explanatory variables but has no 'independent' effect on the dependent variable. Its effect is captured entirely through the explanatory variable, here the government support received after COVID.
Second, to examine the impact of government support on firms' performance, we run the following regression: Where Y is a vector of performance variables, namely: (1) employment and monthly revenue post COVID-19 of firm i in sector j in governorate k; and (2) four dummies that take the value The role of political connections in COVID policy response 1223 of 1 if profits and working hours have declined post-COVID and if firm i in sector j in governorate k had reported laying off any employees post-COVID, or has ever closed down. 15 Gov.SupAC is a dummy variable that takes the value of 1 if the firm has benefited from any post-COVID support measure and zero otherwise. rj and ck are sector and governorate dummies respectively and g is the error term. X is a vector of control variables. In this vector we distinguish between two groups of regressors: (1) 'status variables' or 'innate characteristics' of the firm such as its size, exporting or formality status, its age 16 , its ownership (public or private), its sector and whether it is located in an industrial zone and; (2) behavioral variables that particularly shape the performance and survival of the industrial firm (El-Haddad & Zaki, 2022b, 2023, such as managerial practices, investment in innovation or in worker training and the adoption of advanced technology. Thus, the X vector includes a dummy variable taking the value of 1 if the firm had provided worker training prior to the COVID-19 crisis, a dummy variable that takes the value of 1 if the manager had utilized technology such as computers, the internet, internal information link networks, distributed machine control systems, and quality control systems, a dummy variable that takes the value of 1 if the firm had spent on R&D other than market research surveys, and a dummy variable that takes the value of 1 if the manager had either specified any performance indicators or production targets; or had monitored these performance indicators.
Our analysis is extended to take into consideration firms' heterogeneity. Thus, these extensions examine the differential effect of government support by firm size and age; and by ownership. This helps identify to what extent support has been well allocated and targeted, i.e. whether support goes where it is more effective. Again, given the distribution of firms by age, size and ownership, it is important to identify how skewed the pattern of support is towards politically connected, older, larger and/or public firms; and whether their received support is more or less effective compared to their smaller, younger and private counterparts.

Empirical results
This section presents the empirical findings of the determinants of firms' receipt of post-COVID government support and the effect of that support on a number of performance measures. These include employment, revenues, layoffs, whether the firm had ever closed since the start of the pandemic as well as reported reductions in profits and in working hours. Table 2 presents results from three OLS linear probability regressions, where the dependent variables are the 'overall government support' dummy, the 'financial support' dummy and the 'exemptions and deferments' dummy. 17 The independent variables and instruments are as specified in Equation (1) above. Table S5 presents disaggregated regressions of all detailed government support measures that fall under 'financial government support' and the five distinct debt and tax related support measures that fall under the 'exemptions and deferments' support category. 18 Table 2 shows that overall having received pre-COVID government support, being formal, foreign or located in an industrial zone increases the likelihood for post-COVID 'overall support' and for support under the 'exemptions and deferments' category. Exporting pre-COVID, being large (not SME), being old, public and politically connected additionally increase a firm's chance for receiving post-COVID 'financial support' in particular. These results, confirm the stylized facts presented above. In addition, a firm is more likely to receive overall government support if firms located in the same governorate and the same sector receive a support.

Determinants of government support
The individual regressions for each of the 14 support measures separately (Table S5) show the importance of pre-COVID government support for all measures without exception and political connections for 'financial support' as robust determinants of post-COVID support. Since the systems for pre-COVID support were already in place, identifying firms which had previously received support was straightforward, compared to identifying others that are unknown, unimportant to government, or about which government has limited information. At the same time, previous support, especially financial, is picking up political connections. Political connections are not robustly significant for the 'exemptions and deferments' support category, 19 so possibly these categories of support were targeted to more deserving firms. However, there is still a political element through the influence of connections for support prior to COVID. The shift share variable, while less robust, explains the likelihood of getting government support for land and export drawbacks.
Private firms are consistently less likely to receive 'financial support' from government, other than export drawback (columns 1-10, Table S5). Refunds for incurred export-related expenses (e.g. tariffs incurred on imported inputs, column 10) is the exception since, by definition, only exporters are eligible. Exporters do not become exporters through political favoritism but by being competitive. For all other types of financial support there is evidence of a 'soft budget constraint'. Public firms in Egypt have enjoyed greater protection than their private sector counterparts for years through the provision of cheap state credit and a soft budget constraint. Public companies in crisis are routinely bailed out (El-Haddad, 2015;El-Haddad & Zaki, 2022b). 20 Finally, the tests of Sargan and Basmann of over-identifying restrictions showed that our instruments are valid. All three instruments: political connections, pre-COVID government support and the shift share variable are important determinants for the likelihood of receipt of government support. The role of political connections in COVID policy response 1225

Effect of government support
Each cell of Tables 3 and 4 and of Tables S6-S8 show regression results for which the column heading is the dependent variable. The regressors are the X ijk vector in Equation (2) above, governorate and sector dummies and the support measure. The latter is indicated in the row heading of the tables. Table 3 shows the effect of government support on firm performance for all firms. Tables S6-S8 show same regressions for sub-samples broken down by size, age and ownership respectively. Table 4 shows only aggregated support, namely overall and that broken down by financial and exemptions and deferments government support (i.e. the first three results rows of Tables S6-S8).
The sub sample analysis addresses three research questions: the first is how effective are the various measures, and how is their impact mediated by firm characteristics of interest. The second, is to determine whether the pattern of support presented above as stylized facts is supported by the results. That is, has support gone to where it is most effective? And, finally, how do these results relate to political connections, the missing middle phenomenon and to the discussion on equity and vulnerability?
The results show that government support has helped mitigate the effects of COVID-19, with a significantly larger, favorable impact on smaller, younger and private firms. However, although these firms apparently make better use of government support, they receive a disproportionately smaller share of support.
The first row of Table 3 shows that firms receiving any sort of government support perform significantly better on nearly all outcome measures. A negative coefficient on the adverse impact outcomes in the last four columns means the adverse effect was weaker, so a larger negative coefficient corresponds to better (or less bad) firm performance. Looking at the full sample, the same holds true when government support is broken down into 'financial support' and 'exemptions and deferments' (result rows 2 and 3 in Table 3).

4.2.
1. Effectiveness of government support by size, age and ownership. However, the effect varies by firm characteristics. When we consider the sub-sample estimates in Tables 4(a) and S6, we see that government support has a more significant and larger effect for SMEs compared to large firms. Government support overall significantly improves firm performance for five out of the six outcomes for SMEs, but only one out of six for larger firms (1st result row). Government support has no effect on large firms' employment, it does not reduce their profit loss, or post-COVID falls in working hours or layoffs and there is no effect on large firms' closures since the beginning of the pandemic. In addition, the observed coefficient is higher for small firms than larger firms for every outcome. 21 The same pattern is observed when government support is disaggregated into the two sub-categories in the second and third rows of the table. Only here it is very clear that for large firms 'exemptions and deferments' exert absolutely no effect on all performance indicators, but that 'financial support' in particular in addition to its positive effect on revenues also improves losses in profits. Yet, significance and coefficient size remain considerably favorable with respect to SMEs.
A very similar picture emerges when comparing young versus old firms (Tables 4(b) and S7). There is a significant improvement for four out of six outcomes for young firms, but only three out of six for older firms, and the coefficients are consistently larger for young firms than old ones with respect to both 'financial support' and 'exemptions and deferments'. In particular, the impact of support is remarkable on the adverse impact outcomes 22 of younger firms in contrast to older ones. The same is observed for disaggregated government support measures more or less.
The pattern is stronger still for public versus private firms (Tables 4(c) and S8). Government support, both overall and disaggregated, improves every performance outcome for private firms, but at most three out of six performance indicators for public firms, with absolutely no effect on employment, revenues or the ever-having closed status. This is understandable, as The role of political connections in COVID policy response 1227  is the one mentioned in the column and the independent variables are whether the firm is formal or not, private or not, its age, whether it is located in an industrial zone, whether the firm provides training to its workers or not, uses technology or not, spends on R&D or not, uses good management practices or not, governorate and sector dummies, in addition to the government support presented in each row.
The role of political connections in COVID policy response 1229 support is unlikely to affect rigid public sector employment and more or less relatively stable revenue stream. Again, in the cases when measures are effective for both private and public sector firms, the former coefficients are significantly 23 larger than for their public counterparts. In summary, in answering the second question indicated above, 'financial support' has been incorrectly skewed towards larger, older and public firms ( Table 2) where it has been least effective, thus reinforcing the 'missing middle' phenomenon. In contrast, post-COVID support on 'exemptions and deferments' had been relatively more allocated to favor smaller, younger and private firms post-COVID where it is more effective.

4.2.2.
The most effective government support measures. This section discusses the results in the remaining rows of Tables 3 and S6-S8 in which the types of support are further disaggregated.
Considering the separate interventions of the two sub-categories of support, all debt and taxrelated measures of the 'exemptions and deferments' category have reduced layoffs, which is not the case for any of the 'financial support' interventions. 24 This finding reflects the likely short-term effects of relatively more short-term, crisis induced support measures. On the other hand, 'financial support' measures produce larger positive effects on employment and revenues.
Together with 'export drawback', measures included in the 'exemptions and deferments' category are the six measures most frequently implemented after COVID-19 (Table S9). These six measures reached just under 90% of all firms which received any form of government support. The most effective of these measures in magnitude of effect across the six firm performance indicators is export drawback, followed by allowing delays in loan service payments, and delays in paying taxes. With respect to the least commonly used measures, providing financial support in purchasing and renting land is the most effective followed by the state paying back its old dues to firms. Financial support towards workers insurance is the least effective, perhaps because many workers do not have insurance in the first place. In contrast, there were accumulated government arrears from 2012 of payments due to firms from the export support program. In the wake of the pandemic, about 40 billion Egyptian pounds of arrears were paid within a year and a half (Information Decision & Support Center, 2022), which had indeed favorable effects on the performance of these firms. These payments would fall under the paying back old state dues measure.
In the context of the least commonly used measures of support for purchasing and renting land. As indicated above it is the most effective among its group across the six firm performance indicators. Unequal access to land has been a major constraint to industrial development in developing countries in general (e.g. Altenburg, 2011) as well as in Egypt (Loewe, 2013, El-Haddad, 2016. The government of Egypt is aware of this and has addressed the problem of obtaining industrial land by making land available through the usufruct or purchase right charging only for the cost of utilities. They also allow the Industrial Development Authority to obtain approvals from various authorities on behalf of the investor in less than 20 working days through a process of cooperation and consultation with local and foreign businesses (Information Decision & Support Center, 2022 June 1st). Whether the process of support for obtaining or leasing the land will be transparent, or operate behind a smoke of political favourtism, remains to be seen.

Discussion and conclusion
Our results have four important implications.
First, going back to the two considerations raised in our introduction, our results provide clear answers to the question on vulnerability and equity as well as on effectiveness of the support. We show there is no equity-effectiveness trade-off.
Government support has helped mitigate the effects of COVID-19, with a significantly larger, favorable impact on smaller, younger and private firms. Accordingly, government support is most effective when given to firms which are likely the most vulnerable. However, although these firms make better use of government support, they receive a disproportionately smaller share of it. The main recipients of support are not the most vulnerable and are also not the firms for which support is most effective. This is so since support goes predominately and unjustifiably to politically connected, larger and older firms. This allocation is not justified by economic considerations. 25 In addition, support is given chiefly and unduly to public sector firms rather than the private ones. But our analysis shows support is most effective in improving almost all firm performance indicators for private firms, not public ones. This result is consistent with trends in domestic political economy in recent years, which brings us to the next implication of our results.
Second, the results presented in this paper are relevant to illustrating the 'missing middle' and the role of both the 'soft budget constraint' and 'crony capitalism' in creating the phenomenon in the first place, and how this configuration is relevant to the persistence of Egypt's 'unsocial' social contract.
In Egypt's emerging 'unsocial' social contract, industrial policy has been mediated by statebusiness relations which underpin the deliverables exchanged between the state and businesses in that contract. State-business relations in the country are characterized by the excessive degrees of capture of industrial policy (El-Haddad, 2020b). The 'unsocial' social contract that emerged since the 90s meant that the state used trade, industrial and other economic policies to favour an emerging group of crony capitalists who in turn provided support for the regime. Crony capitalism pads the pockets of the powerful, undermining economic competitiveness and misdirecting resources (Micklethwait and Wooldridge, 2015). Block (2018) describes crony capitalism as an oligarchic market with rule by the twenty or fifty or hundred leading families and business groups. In simple terms, it is a union between capitalist and politicians enabling the former to acquire wealth and the latter to seek and retain power (Pei, 2016). This is what is called the 'unsocial' social contract, where other groups are marginalized (El-Haddad, 2020b).
The situation has only partially changed post COVID-19 which has provided both constraints and opportunities. This paper empirically reveals the persistence of the determinants of the pattern of distribution of government support both pre-and post COVID-19. Political connections are and remain a predominant determinant of receipt of firm-level government support in Egypt 26 (see also recent results from Francis and Kubinec, 2022). Our analysis shows that support received regularly prior to COVID is a chief determinant of post-COVID government support, creating thus a vicious circle of the same recipients of government support. This vicious circle feeds on 'crony capitalism' and a 'soft budget constraint' both of which reinforce 'the missing middle' phenomena in Egypt and contribute to the persistence of the 'unsocial' social contract.
It is important to remember that the 'missing middle' in the Middle East and North Africa Region (MENA), especially in Egypt refers to a concentration of large numbers of very old, large firms on the one hand and very large numbers of very small, younger firms on the other (Schiffbauer et al., 2014, Diwan et al., 2016, Rijkers et al., 2014. Thus, firms in Egypt are much older than their East Asian counterparts for instance. Indeed, Egypt occupies top position in terms of the oldest firms amongst all MENA oil importing countries, with a mean age of 23 years. Egypt's extra-large firms (>600 workers) are twice as old as the group of small firms in our sample.
The third implication of the results is the inefficiency of cronyism. Support instruments typically have eligibility criteria that are automatically applied when thousands of firms are concerned. Allowing institutional discretion introduces inefficiencies. Through bending the eligibility rules, the system of privilege and favor is both costly to implement and also less efficient. So much time is wasted on corruption that is better off used on productive activities.
The final, and perhaps the one positive implication, is that the crisis has presented a chance to make small steps in the right direction. Whilst political connections are principal The role of political connections in COVID policy response 1231 determinants of receipt of firm-level government support in Egypt, they influence 'financial support' measures to a greater degree than they influence debt and tax support measures under 'exemptions and deferments'. By going to the politically connected larger and older firms, 'financial support', with its relatively longer term nature, has been reinforcing the 'missing middle'. However, post-COVID, the crisis has presented a chance for the pattern of support to slowly shift towards the more vulnerable, non-politically connected, smaller, younger and private sector firms through the more frequent use of 'exemptions and deferments' support measures.
There are some caveats to consider when looking at the results of this study. The costs of the various measures has not been taken into account. This study has produced marginal effects for a firm receiving the particular support but has not provided those effects to a dollar unit of that support. A more sophisticated data-set would be warranted to include costing in the assessment and would be a useful extension to the analysis but that is beyond the scope of this paper.
A further potential caveat is that political connections and pre-COVID support are potentially associated with business success through channels other than the receipt of COVIDrelated support. This study has introduced the shift share instrument to partially deal with this problem. Having only cross section data set limits the possibility of finding strictly exogenous variables. Nevertheless, this caveat does not apply to the mostly COVID-induced debt and tax support measures of the 'exemptions and deferments' support category.

Acknowledgment
The contents of this document are the sole responsibility of the authors and do not reflect the position of the BMZ. We are grateful to Dr. Zakaria Othman for undertaking the sampling. Thanks are also due to Marian Adel for superb research assistance and diligence. We are grateful to two anonymous referees to an earlier version of this paper.

Disclosure statement
No potential conflict of interest was reported by the author(s).

Funding
We are thankful for the generous financial support provided by the project 'Stability and Development in the Middle East and North Africa', funded by the German Federal Ministry for Economic Cooperation and Development (BMZ) towards the administration of the 2020/21 Egyptian Industrial Firm Behavior Survey (EIFBS) instrument. Project #:9002101.

Data availability statement
The data set will be made available through ERFs OAMDI portal as of May 2023. Do files are provided upon request.