The Effects of New Firm Formation on Regional Development Over Time the Case of Great Britain

Embed Size (px)

Citation preview

  • 8/10/2019 The Effects of New Firm Formation on Regional Development Over Time the Case of Great Britain

    1/23

    THE EFFECTS OF NEW FIRM FORMATION ON REGIONAL

    DEVELOPMENT OVER TIME: THE CASE OF GREAT BRITAIN

    Pamela Mueller,

    Max Planck Institute of Economics, Germany.

    Andr an S!el,

    Erasmus University Rotterdam, EIM Business & Policy Research, etherlands, and

    Max Planck Institute of Economics, Germany.

    Da"d #$ S!%re&,

    !"ME, #ar$ick Business "chool, University of #ar$ick, U%.

    Paper submitted to Small Business Economics

  • 8/10/2019 The Effects of New Firm Formation on Regional Development Over Time the Case of Great Britain

    2/23

    . I'R()U!'I(

    'his *a*er examines the link +et$een ne$ and small firms on the one hand, and

    economic develo*ment on the other. In *rinci*le it is clear that if a ne$ firm is created

    and it em*loys, either as an o$ner or an em*loyee, individuals $ho formerly $ere not

    em*loyed, then this adds to areate em*loyment in the economy. 'his arithmetic often

    *rovides a su*erficial attraction, *articularly to *oliticians facin hih levels of un-

    em*loyment or under-em*loyment. It suests that raisin ne$ firm formation $ill lead

    su+seuently to additional em*loyment.

    'his *a*er, ho$ever, makes it clear that this outcome is achieved only $hen certain key

    assum*tions are made. 'he first is that it is *ossi+le for *olicy makers to take actions

    $hich do raise ne$ firm formation. / second assum*tion is that macro economic

    circumstances remain unchaned. If this assum*tion is violated 0 such that macro

    economic circumstances +ecome either more or less +enin 0 then this $ill influence

    rates of ne$ firm formation. (n +alance the evidence is that less +enin macro-economic

    conditions lead to a rise in ne$ firm formation. 'he third assum*tion is that it is ne$

    firm formation $hich causes increased em*loyment and not vice versa. (ften, increased

    em*loyment is not the conseuence +ut rather the cause of hiher start-u* rates. 'his is

    +ecause local demand is likely to +e hiher in reions $ith em*loyment increases, and

    therefore such reions are attractive locations to found ne$ firms. /s these t$o

    mechanisms differ reatly in their *olicy im*lications it is of vital im*ortance to

    disentanle the correct direction of causality. 1ourthly, as 1ritsch & Mueller 234456

    sho$, the creation of a ne$ firm has +oth immediate and loner term conseuences. /s

    noted earlier, ne$ firms have an easily identifia+le short term direct effect in creatin

    em*loyment, +ut they also have t$o loner term conseuences. 'he first is that ne$

    firms dis*lace inefficient incum+ents, leadin to 7o+ losses. 8o$ever, a second mediumterm conseuence is that ne$ firms enhance the com*etitiveness of firms that remain in

    +usiness, +y exertin a *o$erful threat u*on incum+ent firms inducin im*roved

    )isney, 8askel and 8edon 234496 find that in the U% +et$een :;4 and ::3 a+out half of *roductivity

    ain $as +ecause of internal factors 0 such as introducin ne$ technoloy and oranisational chanes. 'heremainin half $as +ecause of external factors most nota+ly that the entrants $ere more *roductive than

    those exitin. 8o$ever amonst sinle *lant inde*endent firms almost all the ains $ere attri+uta+le to

    external factors.

    3

  • 8/10/2019 The Effects of New Firm Formation on Regional Development Over Time the Case of Great Britain

    3/23

    *erformance from them. /s 1ritsch and Mueller 234456 sho$, the effect of ne$ entrants

    is therefore threefold< the first effect is to increase em*loyment, the second is to lo$er

    em*loyment and the third is to increase em*loyment. 'he total effect u*on em*loyment

    can therefore +e either *ositive or neative and de*ends u*on the manitude of the three

    elements.

    'his *a*er develo*s the 1ritsch and Mueller 234456 findins for Germany +y a**lyin a

    +roadly similar estimation *rocedure to data for Great Britain. It reconises that the net

    effect of ne$ firm formation on su+seuent em*loyment can +e either *ositive or

    neative. 1urthermore, this study investiates $hether there are circumstances in $hich

    the effects of ne$ firm formation on su+seuent em*loyment chane are more likely to

    +e *ositive or more likely to +e neative. In *articular it assesses $hether *olicies to raise

    ne$ firm formation are more effective in eora*hical areas that already have hih rates

    of ne$ firm formation or $hether they are more effective in lo$ enter*rise areas. 'o test

    this, the *a*er examines the effects of ne$ firm formation on su+seuent em*loyment

    chane +et$een :; and 3449 for Great Britain, makin a distinction +et$een Enland

    on the one hand and "cotland and #ales on the other.

    (ur +road key findin is that, for Great Britain as a $hole and Enland in *articular, the

    effect of ne$ firm formation on su+seuent em*loyment ro$th is *ositive. 8enceraisin ne$ firm formation, iven that this is *ossi+le, $ill lead to su+seuent increased

    em*loyment in this country. 8o$ever, $e do notet the same results for "cotland and

    #ales. 8ere the im*act of increased ne$ firm formation is neative 0 causin lo$er

    su+seuent em*loyment.

    Unfortunately +oth "cotland and #ales have, historically, had lo$ rates of ne$ firm

    formation. Believin this to +e undesira+le, +oth countries im*lemented *olicies,

    inde*endent of those of Enland, $hich have souht to raise ne$ firm formation. 'he

    evidence from this *a*er is that increased rates of ne$ firm formation in "cotland and

    #ales lead to less, rather than more, em*loyment. 'he *a*er therefore raises the uestion

    of $hether it is *ossi+le to have =the $ron ty*e of entre*reneurshi*> and if *u+lic *olicy

    should focus on uantity or uality of ne$ firms. In "cotland and #ales ne$ firm

    9

  • 8/10/2019 The Effects of New Firm Formation on Regional Development Over Time the Case of Great Britain

    4/23

    formation a**ears to lead to fe$er 7o+s $hereas in Enland 0 or at least *arts of Enland

    0 it seems to lead to more 7o+s.

    3. P(?I!@ !('EA'

    'his section discusses the chanin enter*rise *olicy context in Great Britain over the last

    three decades, $hich is vital for understandin the context of the *a*er. #e use the

    cateorisation of *olicy *rovided +y Greene et al. 234456. 'hey identify four se*arate

    *olicy *eriods. 'he first is the :4s $hen enter*rise *olicy 0 defined as *rovidin

    assistance and su**ort for ne$ and small firms 0 hardly existed. 'he :4s is therefore

    vie$ed as a C*olicy offD *eriod.

    'he arrival of Mararet 'hatcher as Prime Minister in :: chaned matters radically. /key tenet of the *hiloso*hy of her overnment $as to move Britain a$ay from $hat she

    sa$ as a de*endency culture 0 defined as one $here individuals relied on the state. 8er

    o+7ective $as for Britain to have an enter*rise culture 0 in $hich individuals relied on

    themselves rather than the state. 'hatcher therefore sa$ enter*rise creation as hihly

    desira+le in itself, +ut also as a direct mechanism for lo$erin unem*loyment levels. 'he

    :;4s can therefore +e considered as a decade characterised +y su**ort for ne$ start-u*

    +usinesses.

    'he ::4s sa$ a shift in this *olicy, a$ay from ne$ firms and to$ards *rovidin su**ort

    for existin small firms $ith ro$th *otential3. 1or GB as a $hole therefore, the mid to

    late ::4s sees considera+ly less em*hasis *laced +y *olicy makers u*on ne$ firms. 'he

    ::4s can therefore +e vie$ed as a decade in $hich the focus $as on smaller firms $ith

    ro$th *otential. 1inally, follo$in the election of a ?a+our overnment in ::, the

    decade of 3444 sees a further chane $ith *olicies +ein focussed more s*ecifically on

    seekin to enhance *roductivity and increase social inclusion. In this *eriod, +oth some

    ty*es of ne$ firms and some ty*es of ro$in firms are su**orted.9

    31or the reasons underlyin this shift see Greene et al 234456.9'hese *olicies $ere formalised in "mall Business "ervice 234456. 'his identified the so-called seven

    *illars< +uildin an enter*rise culture, encourain a dynamic start u* market, +uildin the ca*a+ility forsmall +usiness ro$th, im*rovin access to finance for small +usinesses, encourain more enter*rise in

    disadvantaed communities and under-re*resented rou*s, im*rovin small +usinesses> ex*erience of

    overnment services and develo*in +etter reulation and *olicy

    5

  • 8/10/2019 The Effects of New Firm Formation on Regional Development Over Time the Case of Great Britain

    5/23

    #hilst these *olicies are +ein im*lemented in Great Britain 0 and s*ecifically in

    Enland 0 different *olicies $ere +ein ado*ted in +oth "cotland and #ales, *articularly

    in the ::4s. /t a time $hen GB *olicies focussed u*on Cro$th +usinessesD, "cottish

    *olicy makers +ecame concerned that their ne$ firm formation $as much lo$er than that

    of Enland, as sho$n in 1iure +elo$. In ::9 "cotland launched its Business Birth

    Rate Policy to$ards develo*in +usiness start-u*s and the develo*ment of

    entre*reneurshi* seekin to eliminate the difference +et$een the t$o countries. 'his

    *olicy $as recently revie$ed and an overhaul took *lace. 'he ne$ a**roach $as

    launched in 3443 focusin on three *riorities. 1irstly, innovative hih-ro$th start-u*s

    shall +e encouraed to increase their num+er and value. "econdly, uality volume

    +usiness su**ort services are offered to encourae more *eo*le to start +usinesses.

    1inally, the third focus is on entre*reneurshi* education. (ther tarets are to enerate at

    least 94 ne$ starts that achieve the hihest ro$th $ithin three years, to achieve a 4

    *ercent survival rate for start-u*s over a three-year *eriod, and to increase the num+er of

    start-u*s +y $omen and youn *eo*le.

    Figure 1: VAT Registration per 10,000 inhabitants, 1980-2003

  • 8/10/2019 The Effects of New Firm Formation on Regional Development Over Time the Case of Great Britain

    6/23

    (ver the *eriod :;4 to 3449, ne$ firm formation rates 0 measured as F/' Reistrations

    *er 4,444 inha+itants 0 are consistently lo$er in #ales and "cotland than in Enland

    21iure 6. 'he overall rate of Great Britain is very close to the Enland rate +ecause

    Enland is the numerically dominant country. (n averae a+out :4 *ercent of the F/'

    reistration in Great Britain take *lace in Enland. #ales counts for only a+out four

    *ercent of the reistrations and "cotland conseuently for six *ercent. 1iure also

    sho$s that there is a similar *attern over time in Enland, "cotland and #ales, $ith ne$

    firm formation rates risin in all three countries at least until :;:, then fallin until

    ::5-: and su+seuently +ein +roadly sta+le. 8ence, althouh the ne$ firm formation

    rates reatly chaned over time the distri+ution +et$een the three reions is very constant

    over the *eriod :;4 to 3449. 'he reason for the stee* decrease at the +einnin of the

    ::4s and in ::5 is an increase of the F/' reistration threshold of more than

    94 *ercent and 34 *ercent, res*ectively.

    /lthouh the introduction of the "cottish Business Birth Rate Policy in ::9 seems to

    have resulted in realisin hiher ne$ firm formation rates than #ales 21iure 6 it had

    little o+vious im*act u*on ne$ firm formation and it certainly did not a**ear to close the

    a* $ith Enland. In s*ite of the "cottish ex*erience, #ales also souht to raise ne$

    firm formation. It introduced its Entre*reneurial /ction Plan 2E/P6 in 344 +ut any

    effect is difficult to identify iven its short duration. #ales> Entre*reneurial /ction Plan

    is +uilt around three key o+7ectives. 'he first taret is to encourae more *eo*le to start

    or ro$ a firm +y creatin a reater a$areness of the o**ortunities and +enefits of

    entre*reneurshi* in order to develo* an entre*reneurial culture. "econdly, the focus is on

    sustaina+le start-u*s and to increase their num+er. 'hirdly, the *roram is ouht to

    increase the num+er of start-u*s $ith ro$th *otential and to create em*loyment.

    9. ?I'ER/'URE REFIE#

    'he laed effects of ne$ firm formation on em*loyment ro$th already +ecame

    a**arent in the study of /udretsch and 1ritsch 234436. 'hey found that em*loyment

    ro$th in the ::4s could not +e ex*lained +y start-u* rates in the ::4s +ut rather +y

    start-u* rates of the :;4s in #est Germany. 1ritsch and Mueller>s $ork on Germany,

  • 8/10/2019 The Effects of New Firm Formation on Regional Development Over Time the Case of Great Britain

    7/23

    referred to a+ove, formally sho$ed the im*ortance of time las in examinin the

    relationshi* +et$een ne$ firm formation and em*loyment chane. 'hey theorised that

    ne$ firm formation had three effects reflected in 1iure 3 +elo$.

    'he first effect $as the short term directeffect of em*loyment created +y the ne$ firm.

    'he second $as the displaceenteffect of the entrant $hich causes some existin firms

    to o out of +usiness, hence incurrin 7o+ losses. 1inally, there is a third effect of the

    entrant $hich is to stimulate +etter *erformance from survivin firms. 'his is called the

    inducedeffect.

    /**lyin this theory to German data 1ritsch and Mueller 234456 find su**ort for this

    three *hase model. 'hey concluded that the *eak of *ositive im*act of ne$ firms on

    reional economic develo*ment $as reached a+out eiht years after entry, and

    disa**eared after a+out a decade. Usin a +roadly similar a**roach, van "tel and "torey

    234456 examined Great Britain for the *eriod :; to ::;. 'heir conclusions $ere

    similar to those of 1ritsch and Mueller. 'hey found the em*loyment im*act of ne$ firm

    formation $as maximised in year five +ut had declined to Hero +y year nine.5

    5/ disadvantae of their exercises $as that they restricted the /lmon la *olynomial to +e of second deree

    $hile 1ritsch and Mueller sho$ed that a third deree *olynomial may actually +e more realistic.

  • 8/10/2019 The Effects of New Firm Formation on Regional Development Over Time the Case of Great Britain

    8/23

    Figure 2: !cheatic e""ects o" ne# "ir "oration on eplo$ent change %according to

    Fritsch and &ueller, 200'(

    8o$ever, it is Fan "tel and "torey>s findins for =un-entre*reneurial areas> such as orth-

    East Enland and "cotland, $hich is of reatest interest in the context of the current

    *a*er. 8ere they found that althouh there $as, in the ::4s evidence that for Enland as

    a $hole, ne$ firm formation led to additional em*loyment, this $as not the case for all

    eora*hical areas. In areas $ith lo$ rates of ne$ firm formation such as "cotland and

    orth East Enland, the overall im*act of ne$ firm formation on em*loyment chane

    $as neative. 'his im*lied that *olicies to raise ne$ firm formation in lo$ enter*rise

    areas could +e counter-*roductive in terms of em*loyment chane. Rather than creatin

    ne$ em*loyment o**ortunities, the ne$ firms dis*lace existin firms leadin to

    em*loyment losses or those ne$ firms exit the market shortly after entry there+y

    deradin the em*loyment 7ust created.

    5. )/'/ "(UR!E"

    'he Fan "tel and "torey 234456 analysis ended in ::;, and so there $as insufficient time

    to examine the im*act of the "cottish Business Birth Rate Policy. 'he #ales

    Entre*reneurial /ction Plan 2E/P6 $as not im*lemented until 344 after the data set hadended. 'he current *a*er extends the time *eriod examined u* until 3449, $hich does

    ena+le the "cottish *olicy to +e more effectively revie$ed. 'he E/P is ho$ever too

    recent for any of its effects to +e included.

    'his *a*er formally meres the methodoloy of Fan "tel and "torey, $ith that of 1ritsch

    and Mueller. It has the s*ecific o+7ective of examinin the im*act of *olicy in the t$o

    lo$ enter*rise countries of "cotland and #ales, in com*arison $ith hih enter*rise,

    Enland. /s in the Fan "tel and "torey 234456 *a*er, data on start-u*s is taken from the

    U% "mall Business "ervice. "tart-u*s have to exceed a turnover threshold in order to +e

    reistered for F/' *ur*oses, for instance, a threshold of ,444.44 had to +e exceeded

    in 3449. Businesses +elo$ this threshold may also choose to reister, +ut this is o*tional

    and unlikely. #hile a+out 5,444 start-u*s 2includin ariculture6 reistered in

    ;

  • 8/10/2019 The Effects of New Firm Formation on Regional Development Over Time the Case of Great Britain

    9/23

    Enland, a+out ,444 ne$ firms reistered in #ales and 3,444 in "cotland durin the

    year 3449.

    )ata on em*loyment is taken from the !ensus of Em*loyment and, from :: on$ards,

    the /nnual Em*loyment "urvey. )ata are su**lied +y omis $ith the self-em*loyed and

    un*aid family $orkers +ein excluded from the data. 1urthermore, follo$in Fan "tel

    and "torey, hourly $aes and *o*ulation density $ere included as control varia+les.

    #ae data are derived from the e$ Earnins "urvey Panel 2E"P6 )ataset and $ere

    chaned into constant *rices 3449. )ata on *o*ulation density $ere taken from the (ffice

    for ational "tatistics.

    'he unit of analysis is 4 British reions 0 5 Enlish !ounties, 5 #elsh Reions and 4

    "cottish ?ocal /uthority Reions over the *eriod :;-3449. Because of missin

    2em*loyment6 data, the reion (rkneyJ"hetlandJ#estern Isles had to +e excluded,

    therefore eneratin : o+servations. )ifferent reional and sectoral classifications in the

    oriinal data files meant some linkin o*erations $ere *erformed to ensure uniformity for

    :;403449 2see Fan "tel and "torey, 3445 for details6. 'he aricultural sector is

    excluded as it is fundamentally different from the rest of the economy havin, durin this

    *eriod, exce*tionally lo$ start-u* and death rates.

    . )/'/ //?@"I"

    'he reions in Enland and the reions in "cotland and #ales clearly sho$ the same

    *attern reardin the correlation of ne$ firm formation over time 2see fiure 96. 1irstly,

    there is a stron correlation in the short term +et$een start-u* rates, +ut this correlation

    $eakens over time. "econdly, there is a stron variation over s*ace, $hile some reions

    count for only four start-u*s *er 44 em*loyees, other reions had even more than

    reistrations *er 44 em*loyees. Es*ecially the Enlish reions ex*erience hiher start-

    u* rates. Most "cottish or #elsh reions cluster around lo$er values. 1iure 9 *lots the

    correlation of ne$ firm formation for t$o time *eriods. 'he first is the relationshi*

    +et$een start-u*s in t and in t-and the second is +et$een t and t-. / distinction is made

    +et$een the Enland, and "cotland and #ales data.

    :

  • 8/10/2019 The Effects of New Firm Formation on Regional Development Over Time the Case of Great Britain

    10/23

    Figure 3a: )orrelation o" ne# "ir "oration rates o*er tie in +ngland %per 1,000

    eplo$ees(

    Figure 3b: )orrelation o" ne# "ir "oration rates o*er tie in ales and !cotland %per1,000 eplo$ees(

    'he results of 'a+le underline the *ersistency of ne$ firm formation activity over time

    at the reional level. 'he start-u* rate in *eriod t heavily de*ends on the start-u* rate in

    the *revious year and is also sinificantly determined +y ne$ firm formation activity five,

    ten or years ao. 'he +eta-coefficients indicate the stron *attern of *ath de*endency

    $hich $eakens over time. More than ;4 *ercent of the variation of the start-u* rate in t

    can +e ex*lained +y ne$ firm formation activity one, five, ten and years ao.

    4

  • 8/10/2019 The Effects of New Firm Formation on Regional Development Over Time the Case of Great Britain

    11/23

    Table 1: ath dependenc$ o" ne# "ir "oration acti*it$ o*er tie

    Start-up ratet

    Start-up ratet-1 0.935**(95.04)

    0.599**(15.62)

    Start-up ratet-5 0.548**(21.33)

    0.294**(9.88)

    0.600**(21.72)

    0.682**(28.54)

    Start-up ratet-10 0.503**(16.10)

    -0.011(0.41)

    0.205**(7.28)

    0.221**(9.27)

    0.600**(19.71)

    Start-up ratet-15 0.643**(18.22)

    0.142**(7.46)

    0.249**(11.40)

    0.337**(11.07)

    R-adjusted 0.8744 0.2996 0.2522 0.4126 0.8943 0.8394 0.6376 0.6781

    F-Vaue 9032.16 454.94 259.27 331.91 996.77 821.39 674.78 497.05

    !"ser#at$%&s 1298 1062 767 472 472 472 767 472

    ** significant at 1%-level, t-value in parentheses, beta-coefficients.

    . 1(RM/??@ M()E??IG 'IME-?/G"

    'o formally address the issue of time-las, the first ste* is to estimate a model relatin

    em*loyment chane over a t$o year *eriod 2+et$een t and tK36 to start-u* rates in year t

    and each of the *recedin eiht years. 'he results for Great Britain are sho$n in 'a+le 3,

    toether $ith measures of *o*ulation density, $aes, laed em*loyment chaneand a

    term to address s*atial auto correlation 2see e.. %ee+le et al., ::9, or /nselin, :;;6.

    'he results are sho$n ra*hically in 1iure 5. 'he smoothed line o+tained in 1iure 5 is

    dra$n on the +asis of the so-called restricted *arameters in 'a+le 3. 'hese restrictions are

    necessary +ecause the initial reression suffers from multicollinearity caused +y the hih

    correlations over time of the start-u* rates 2see 'a+le 6. 'hese *ro+lems of

    multicollinearity are avoided usin the /lmon la method. Basically this method im*oses

    restrictions on the *arameters of the start-u* rates in such a $ay that the estimated

    coefficients of the start-u* rates are a function of the la lenth. By su+stitutin these

    'his varia+le is included to control for the reversed causality issue discussed in the introduction. 'heconce*t of usin laed de*endent varia+les to test for the direction of causality is kno$n in the

    econometric literature as testin for Graner-causality. 'he Graner 2::6 a**roach to the uestion of

    $hether x causes y is to see ho$ much of the current y can +e ex*lained +y *ast values of y and then to see

    $hether addin laed values of x can im*rove the ex*lanation. y is said to +e Graner-caused +y x if xhel*s in the *rediction of y, or euivalently if the coefficients on the laed x>s are statistically sinificant.

    "ee /udretsch et al. 23446, for an a**lication of these Graner tests in the context of modellin the relation

    +et$een self-em*loyment and unem*loyment for 39 (E!) countries.

  • 8/10/2019 The Effects of New Firm Formation on Regional Development Over Time the Case of Great Britain

    12/23

    restrictions +ack in the oriinal euation one arrives at a more com*act model $hich can

    +e estimated $ithout *ro+lems of multicollinearity. #e refer to "te$art 2::, **. ;4-

    ;36 for a eneral descri*tion of the /lmon method.

    It is clear from 'a+le 3 usin +oth a fixed effects model and a 8u+er-#hite Ro+ust

    Model that there are three different im*acts. 'he initial effect is *ositive in year t +ut

    +ecomes neative in t- and is sinificantly neative in t-9. Both models sho$ that, in

    year t-, effects +ecome sinificantly *ositive. Gra*hically this may +e seen +y

    com*arin the areas under and a+ove the horiHontal line. 'he area +elo$ the horiHontal

    line in 1iure 5 can +e considered as the neative em*loyment effects of ne$ firm

    formation. 'he *ositive effects are sho$n a+ove the horiHontal line 0 the *eriod of u* to

    year and the *eriod from year 5 to year ;. 'he net effects $hich are *ositive for Great

    Britain are clearly sho$n in 1iure 5. !oncernin our control varia+les, $e find *ositive

    effects of hourly $aes and s*atial autocorrelation, and neative effects of laed

    em*loyment ro$th 2indicatin +usiness cycle effects6 and *o*ulation density. 'his last

    varia+le is meant to control for reional characteristics like housin *rices, ualified

    la+our, local demand, and kno$lede. ote that the coefficient is not sinificant usin the

    fixed effect estimator since reional differences $hich are time-*ersistent 2such as

    *o*ulation density6 are already accounted for in the fixed effect.

    /s the fixed effects method effectively considers the effects over time only, and the *ur*ose of our *a*eris to ex*licitly examine the short and lon run effects over time, $e mainly focus on the fixed effects

    results in our discussion. 8o$ever, the 8u+er #hite results are useful for ro+ustness test *ur*oses.#hen usin fixed effects $e realise that inclusion of a laed de*endent varia+le miht lead to a +ias in

    the estimation results. 8o$ever, com*arin the coefficient for the laed de*endent of the 1E reression$ith that of the 8u+er #hite reression $e see that these are very similar. 'herefore $e arue that this +ias

    is, at the most, small. Because the effect of the laed de*endent is very sinificant $e choose to include it

    in our model, des*ite the disadvantae mentioned.

    3

  • 8/10/2019 The Effects of New Firm Formation on Regional Development Over Time the Case of Great Britain

    13/23

    Table 2: .ong-ter e""ects o" ne# "ir "oration acti*it$ on eplo$ent change

    '%-ear re$%&a e+p%+e&t ,a&e ()

    F$/ed ee,t est$+at%r !S est$+at%r (u"er $te)

    u&-restr$,ted3rd %rderp%&%+$a u&-restr$,ted

    3rd %rderp%&%+$a

    Start-up rate (t) 0.881**(3.81)

    0.828 0.732**(4.36)

    0.678

    Start-up rate (t-1) -0.492(1.78)

    -0.205 -0.529**(2.78)

    -0.267

    Start-up rate (t-2) 0.027(0.10)

    -0.569 -0.083(0.48)

    -0.576

    Start-up rate (t-3) -0.878**(3.24)

    -0.468 -0.823**(3.73)

    -0.451

    Start-up rate (t-4) -0.206(0.75)

    -0.111 -0.141(0.54)

    -0.089

    Start-up rate (t-5) 0.508

    (1.83)

    0.297 0.499

    (1.82)

    0.308

    Start-up rate (t-6) 0.732**(2.66)

    0.549 0.729**(2.82)

    0.542

    Start-up rate (t-7) -0.026(0.10)

    0.440 -0.105(0.46)

    0.412

    Start-up rate (t-8) 0.041(0.22)

    -0.238 0.009(0.05)

    -0.281

    %puat$%& de&s$t 0.005(0.54)

    0.006(0.60)

    -0.001**(4.37)

    -0.001**(4.37)

    %ur aes(,%&sta&t pr$,es)

    1.108**(2.64)

    1.072**(2.57)

    0.546**(3.51)

    0.558**(3.74)

    2-ear e+p%-+e&t ,a&e (t-2)

    -0.333**(9.35)

    -0.326**(9.44)

    -0.259**(5.88)

    -0.257**(5.99)

    Spat$aaut%,%rreat$%&

    0.590**(8.88)

    0.568**(8.54)

    0.542**(6.62)

    0.524**(6.54)

    %&sta&t 14.257*(2.12)

    -13.716*(2.04)

    -3.700**(3.18)

    -3.725**(3.28)

    R-adjusted 0.2946 0.2899 0.3167 0.3138F-Vaue 30.07 47.33 21.00 33.27%-$e$%%d -2059.80 -2065.11 -2098.85 -2102.97

    !"ser#at$%&s 767 767 767 767

    9

  • 8/10/2019 The Effects of New Firm Formation on Regional Development Over Time the Case of Great Britain

    14/23

    -0.6

    -0.3

    0.0

    0.3

    0.6

    0.9

    7+pa,t%0&e(0$r+0

    %r+

    at$%&

    %&e+p%)+e&t,-a&

    *e

    0 1 2 3 4 5 6 7 8

    a*s (ears)

    Figure ': /pact o" ne# "ir "oration on eplo$ent change "or ritish regions %"ied

    e""ects estiator(

    )oparing +ngland #ith ales and !cotland

    In this section a distinction is made +et$een the results for Enland, and those for #ales

    and "cotland. /n interaction dummy indicates the ne$ firm formation activity inEnland in com*arison to ne$ firm formation in "cotland and #ales. 'he results are

    sho$n in ta+le 9 and *lotted in fiure . Unsur*risinly, the results for Enland are very

    similar to the overall results of Great Britain. 'he net im*act of ne$ firm formation is

    overall *ositive in Enland. In *articular, the im*act of the third *hase defined as the

    com*etitiveness *hase 0 is sinificantly larer than the second *hase 0 defined as the

    dis*lacement *hase.

    5

  • 8/10/2019 The Effects of New Firm Formation on Regional Development Over Time the Case of Great Britain

    15/23

    Table 3: .ong-ter e""ects o" ne# "ir "oration acti*it$ on eplo$ent change

    +ngland *s !cotland and ales

    F$/ed ee,ts est$+at%r !S est$+at%r (u"er $te)u&-restr$,ted 3rd %rder p%&%+$a u&-restr$,ted 3rd %rder p%&%+$a

    England:

    Start-up rate (t) 0.915**(3.56) 0.814 0.636**(3.32) 0.596

    Start-up rate (t-1) -0.572(1.81)

    -0.202 -0.581**(2.83)

    -0.313

    Start-up rate (t-2) -0.052(0.17)

    -0.559 -0.203(1.07)

    -0.593

    Start-up rate (t-3) -0.598*(1.93)

    -0.458 -0.528*(2.34)

    -0.445

    Start-up rate (t-4) -0.304(0.97)

    -0.098 -0.261(0.96)

    -0.007

    Start-up rate (t-5) 0.357(1.12)

    0.318 0.368(1.46)

    0.339

    Start-up rate (t-6) 0.830**(2.56)

    0.592 0.789**(2.82)

    0.575

    Start-up rate (t-7) 0.342(1.08)

    0.521 0.293(1.09)

    0.442

    Start-up rate (t-8) -0.077

    (0.35)

    -0.094 -0.236

    (1.27)

    -0.259

    Scotland and Wales:Start-up rate (t) 1.344**

    (2.94)1.095 1.510**

    (4.66)1.102

    Start-up rate (t-1) -0.293(0.52)

    -0.145 -0.378(1.02)

    -0.037

    Start-up rate (t-2) 0.028(0.05)

    -0.613 -0.034(0.10)

    -0.498

    Start-up rate (t-3) -1.878**(3.40)

    -0.551 -1.790**(4.06)

    -0.482

    Start-up rate (t-4) 0.323(0.59)

    -0.204 0.515(0.84)

    -0.193

    Start-up rate (t-5) 0.981(1.78)

    0.186 0.937(1.41)

    0.166

    Start-up rate (t-6) 0.351(0.67)

    0.376 0.377(0.80)

    0.392

    Start-up rate (t-7) -1.239*(2.48) 0.121 -1.395**(2.70) 0.282

    Start-up rate (t-8) 0.126(0.32)

    -0.821 0.674*(0.27)

    -0.368

    %puat$%& de&s$t 0.003(0.27)

    0.003(0.33)

    -0.001**(4.19)

    -0.001**(4.33)

    %ur aes(,%&sta&t pr$,es)

    1.126**(3.00)

    1.148**(2.74)

    0.554**(3.79)

    0.557**(3.88)

    2-ear e+p%+e&t,a&e (t-2)

    -0.335**(9.20)

    -0.346**(9.91)

    -0.252**(5.78)

    -0.267**(6.06)

    Spat$a aut%,%rreat$%& 0.524**(7.57)

    0.509**(7.39)

    0.475**(5.05)

    0.479**(5.45)

    %&sta&t 14.733*(2.19)

    -13.491*(2.00)

    -3.898**(3.25)

    -3.805**(3.17)

    R-adjusted 0.3078 0.2978 0.3274 0.3179F-Vaue 19.12 32.91 35.41 31.93%-$e$%%d -2047.55 -2058.60 -2088.17 -2098.68

    !"ser#at$%&s 767 767 767 767

    'here is a strikin difference +et$een results for Enland and those for #ales and

    "cotland 2ta+le 96. /lthouh the +asic *attern of three *hases is the same, the #ales and

    "cotland result sho$s almost no em*loyment ain in the com*etitiveness *hase. #ales

  • 8/10/2019 The Effects of New Firm Formation on Regional Development Over Time the Case of Great Britain

    16/23

    and "cotland therefore are characterised +y a relatively larer *hase and a loner

    neative *hase 3 in com*arison $ith Enland, and an almost non-existent *hase 9. 'his

    im*lies that the characteristics of ne$ firms in #ales and "cotland differ shar*ly from

    those in Enland.

    "o, on the +asis of 1iure $e o+serve differences +et$een Enland on the one hand and

    "cotlandJ#ales on the other. But are these differences sinificantL 'o test this $e carry

    out a likelihood ratio test 2?R test6 com*arin the models includin the interaction

    dummies from 'a+le 9 $ith the eneral model from 'a+le 3. 'here are t$o cases.First,

    there are the +asic 2unrestricted6 estimations. In terms of the ?R test the lolikelihood

    value for the unrestricted model is -345. and that of the restricted model 2i.e.

    =restrictin> the effects of Enland and "cotlandJ#ales to +e the same6 -34:.;4. 'he

    corres*ondin ?R test statistic 2takin t$o times the difference6 thus euals 35.. 'he

    critical value of the chi-suared distri+ution $ith : derees of freedom 2there are :

    restrictions corres*ondin to the : interaction dummies6 is .:3 2at sinificance

    level6. 8ence the null hy*othesis of valid restrictions is re7ected. !econd, there are the

    restricted 29rdorder *olynomial6 estimations. 8ere the test statistic amounts to a value of

    9.4 2N3O234.-34;.66 and the critical value is :. 2note that there are only four

    restrictions here corres*ondin to the four additional *olynomial terms for "cotland and

    #ales6. "o aain the null hy*othesis is re7ected. 8ence the difference in im*act +et$een

    Enland and "cotlandJ#ales is statistically sinificant.

  • 8/10/2019 The Effects of New Firm Formation on Regional Development Over Time the Case of Great Britain

    17/23

    Figure 4: /pact o" ne# "ir "oration on eplo$ent change +ngland *s !cotlandand ales %"ied e""ects estiator(

    )oparing lo# and high entrepreneurial areas

    'he analysis a+ove makes it clear that the characteristics of ne$ firms in #ales and

    "cotland differ from those in Enland. #e arued that this is related to the differences instart-u* rates, i.e. on averae relatively hih start-u* rates in Enland and lo$er rates in

    #ales and "cotland 2see 1iure 6. In this section $e test this arument more formally.

    In order to analyse the differences +et$een lo$ and hih entre*reneurial areas in more

    detail $e include an interaction term classifyin reions as lo$ entre*reneurial areas.

    Reions are classified lo$ entre*reneurial if they +elon to the lo$er 3 *ercentile

    reardin their averae start-u* rate. Results are in 'a+le 5 and 1iure . /lthouh the

    lo$ entre*reneurial reions seem to ex*erience a stron direct effect;

    they are

    characteriHed +y a lare dis*lacement effect $hich lasts much loner com*ared to the

    hihest of reions in terms of start-u* activity in Great Britain. 1urthermore these

    ;Usin data for the etherlands, Fan "tel and "uddle 23446 demonstrate that there is a daner of

    overestimatin the direct effect $hen a**lyin the /lmon method. 'his is due to reversed causality effects.

  • 8/10/2019 The Effects of New Firm Formation on Regional Development Over Time the Case of Great Britain

    18/23

    reions have a very lo$ inductive effect $hich is also much shorter in time. 'he results

    clearly indicate a neative net effect for lo$ entre*reneurial areas.

    /ain, com*arin lolikelihood values from 'a+les 3 and 5, $e can formally test $hether

    the lo$ entre*reneurial areas differ from the other areas. 'he ?R test statistics for the t$o

    cases identified a+ove are no$ 94.; and 3.4. !om*ared to our earlier exercise

    2"cotlandJ#ales vs. Enland6 these values are hiher 2the ?R test statistics $ere 35. and

    9.4 in the earlier exercise, res*ectively6 indicatin that the null hy*othesis of an eual

    effect in lo$ and hih entre*reneurial areas is re7ected even more stronly. 'his is

    illustrated in 1iure $hich com+ines the *atterns found in 1iures and .

    ;

  • 8/10/2019 The Effects of New Firm Formation on Regional Development Over Time the Case of Great Britain

    19/23

    Table ': .ong-ter e""ects o" ne# "ir "oration acti*it$ on eplo$ent change lo#

    entrepreneurial areas

    '%-ear re$%&a e+p%+e&t ,a&e ()

    F$/ed ee,ts est$+at%r !S est$+at%r (u"er $te)

    u&-restr$,ted 3rd %rder p%&%+$a u&-restr$,ted 3rd %rder p%&%+$aControl group (!"

    percentile#:Start-up rate (t) 0.751**

    (3.02)0.718 0.537**

    (2.93)0.540

    Start-up rate (t-1) -0.521(1.72)

    -0.211 -0.551**(2.87)

    -0.298

    Start-up rate (t-2) 0.105(0.35)

    -0.524 0.005(0.03)

    -0.547

    Start-up rate (t-3) -0.710*(2.36)

    -0.412 -0.665**(2.61)

    -0.397

    Start-up rate (t-4) -0.432(1.42)

    -0.066 -0.383(1.69)

    -0.038

    Start-up rate (t-5) 0.798**(2.58)

    0.321 0.799**(2.75)

    0.340

    Start-up rate (t-6) 0.590*

    (1.89)

    0.560 0.568*

    (1.84)

    0.547

    Start-up rate (t-7) 0.054(0.18)

    0.459 0.006(0.02)

    0.395

    Start-up rate (t-8) 0.086(0.41)

    -0.174 -0.086(0.41)

    -0.308

    $o entrepreneurialareas (&'!" percentile#Start-up rate (t) 1.354**

    (2.72)1.452 1.756**

    (5.06)1.557

    Start-up rate (t-1) -0.117(0.18)

    -0.295 -0.233(0.39)

    -0.064

    Start-up rate (t-2) -0.429(0.71)

    -0.895 -0.483(1.12)

    -0.774

    Start-up rate (t-3) -1.856**(3.10)

    -0.901 -1.749**(4.11)

    -0.825

    Start-up rate (t-4) 0.272(0.46)

    -0.505 0.538(0.84)

    -0.472

    Start-up rate (t-5) -0.420(0.71)

    0.011 -0.415(0.84)

    0.034

    Start-up rate (t-6) 0.943(1.67)

    0.365 0.975(1.83)

    0.438

    Start-up rate (t-7) -0.436(0.76)

    0.275 -0.530(0.93)

    0.488

    Start-up rate (t-8) -0.163(0.37)

    -0.541 0.493(1.62)

    -0.068

    %puat$%& de&s$t 0.010(0.97)

    0.011(1.04)

    -0.001**(4.43)

    -0.001**(4.60)

    %ur aes(,%&sta&t pr$,es)

    1.003*(2.36)

    0.958*(2.27)

    0.566**(3.72)

    0.573**(3.90)

    2-ear e+p%+e&t,a&e (t-2)

    -0.350**(9.75)

    -0.343**(9.93)

    -0.266**(6.04)

    -0.264**(6.23)

    Spat$a aut%,%rreat$%& 0.558**(7.96)

    0.533**(7.64)

    0.510**(5.42)

    0.496**(5.47)

    %&sta&t -14.173*(2.13)

    -13.633*(2.05)

    -3.659**(2.92)

    -3.569**(2.85)

    R-adj. adjusted 0.3135 0.3096 0.3274 0.3247F-Vaue 19.54 34.46 33.28 38.27%-$e$%%d -2044.38 -2052.10 -2088.14 -2094.80!"ser#at$%&s 767 767 767 767

    :

  • 8/10/2019 The Effects of New Firm Formation on Regional Development Over Time the Case of Great Britain

    20/23

    Figure 5: /pact o" ne# "ir "oration on eplo$ent change in lo# entrepreneurialareas %belo# 24 percentile(

    Figure 6: /pact o" ne# "ir "oration on eplo$ent change in !cotland and ales*s the lo# entrepreneurial areas %belo# 247(

    34

  • 8/10/2019 The Effects of New Firm Formation on Regional Development Over Time the Case of Great Britain

    21/23

    . IMP?I!/'I("

    'his *a*er has em*hasised that there is an immediate *ositive im*act of ne$ firm

    formation on em*loyment. 8o$ever, it also em*hasises that this is not the only 0 and

    indeed not the most im*ortant 0 conseuence. !onfirmin the earlier results of 1ritsch

    and Mueller 234456, it finds that the im*act is in three discrete *hases. 'he first and

    immediate im*act is in terms of 7o+ creation, a second im*act $hich is neative in terms

    of entrants dis*lacin inefficient incum+ent firms, and a third im*act in terms of

    enhanced com*etitiveness. 'he full effects of this im*act are felt over a decade.

    #here the *a*er moves for$ard the $ork of 1ritsch and Mueller is to em*hasise that the

    im*act on em*loyment of ne$ firm formation varies s*atially $ithin Great Britain. It

    sho$s +oth informally +y *resentin *ictures and formally +y a**lyin statistical tests,

    that the im*act is sinificantly reater in Enland 0 an area $ith hih rates of ne$ firm

    formation, than it is in #ales and "cotland, $hich have lo$ rates of ne$ firm formation.

    Indeed, its strikin findin is that in the lo$ enter*rise countries of "cotland and #ales,

    the im*act of ne$ firm formation is overall neative. "*ecifically, investiatin the

    em*loyment effects in lo$ entre*reneurial reions in Great Britain confirms the overall

    neative net-effect of ne$ firm formation on em*loyment chane in lo$ enter*rise areas.

    /lthouh these areas seem to ex*erience a stroner direct effect the ne$ firms do notseem to facilitate the induced 2lon term6 effect.

    'his has *otentially sinificant conseuences for *olicy since +oth "cotland and #ales

    have im*lemented *olicies to raise ne$ firm formation in their countries. In "cotland,

    the =Business Birth Rate> *olicy im*lemented +y "cottish Enter*rise incurred 54m of

    tax *ayers> money after ::9, until its a+olition in 344. 'he official revie$ of this

    *olicy initiative $as that it failed either to raise ne$ firm formation in a+solute terms or

    to close the a* +et$een "cotland and Enland.:

    (ur results in this *a*er are even more dammin. 'hey suest that, even if "cottish

    Enter*rise had +een a+le to raise ne$ firm formation, this $ould have in fact led to lo$er,

    :1raser of /llander 23446

    3

  • 8/10/2019 The Effects of New Firm Formation on Regional Development Over Time the Case of Great Britain

    22/23

    rather than hiher, su+seuent em*loyment ro$th than $as actually achieved. In effect it

    $as a +lessin that the *olicy, even thouh it cost 54m, $as ineffective.

    (ur concern is that the *olicies to raise ne$ firm formation $ere then introduced in

    #ales in 344, and iven #ales and "cotland a**ear +roadly similar, even if #ales is

    successful in raisin ne$ firm formation this may not necessarily lead to more ne$ 7o+s

    +ein created in #ales.

    Unfortunately, $e are not a+le to assess the uality of the ne$ly founded firms and to

    identify the rightand #rongty*e of entre*reneurshi*. In order to do so future research

    needs to evaluate data on hih ro$th firms and entre*reneurshi* in industries $ith

    ro$th *otential.

    33

  • 8/10/2019 The Effects of New Firm Formation on Regional Development Over Time the Case of Great Britain

    23/23

    RE1ERE!E"

    /nselin, ?. 2:;;6, !patial +conoetrics: &ethods and &odels, )ordrecht< %lu$er/cademic Pu+lishers.

    /udretsch, ).B., M./. !arree, /.. van "tel and /.R. 'hurik 23446, =)oes "elf-

    Em*loyment Reduce Unem*loymentL>, Max Planck Institute )iscussion Pa*er Q4-344, ena< Max Planck Institute of Economics.

    )isney, R, 8askel and 8eden, @ 234496, CRestructurin and Productivity ro$th in the

    U%D,+conoic ournal,uly, Fol. 9, o. 5;:, **. -:5.

    1raser of /llander Institute 23446, Crooting usiness !tart ups: A e# !trategic

    ForulaD, University of "trathclyde, Glaso$

    1ritsch, M. and Mueller, P, 234456,C'he Effects of ne$ +usiness formation on Reional)evelo*ment over 'imeD,Regional !tudies, Fol. 9;, o , ovem+er, ** :-:.

    Graner, !.#.. 2::6, Investiatin !ausal Relations +y Econometric Models and!ross-"*ectral Methods,+conoetrica'(, 535-59;.

    Greene, 1.., Mole, %.1. and "torey, ).. 234456, C)oes more mean $orseL 'hree decades

    of Enter*rise *olicy in the 'ees FalleyD, rban !tudies, Fol. 5, o , **. 34-33;

    %ee+le, )., ". #alker and M. Ro+son 2::96, e$ 1irm 1ormation and "mall Business

    Gro$th in the United %indom< "*atial and 'em*oral Fariations and )eterminants,

    Research "eries, o. , "heffield, U%< Em*loyment )e*artment.

    "mall Business "ervice 234456, CA go*ernent action plan "or sall businessesD,

    )e*artment of 'rade and Industry, ?ondon.

    "te$art, . 2::6,+conoetrics, e$ @ork< Phili* /llan.

    Fan "tel, /.. and "torey, ).. 234456, C'he ?ink +et$een firm +irths and o+ !reation< Isthere a U*as 'ree EffectLD,Regional !tudies, Fol. 9;, o ;, ovem+er, ** ;:9-:4.

    Fan "tel, /.. and %. "uddle 23446, C'he Im*act of e$ 1irm 1ormation on Reional

    )evelo*ment in the etherlandsD, !all usiness +conoics, *resent issue.